-
Notifications
You must be signed in to change notification settings - Fork 0
Expand file tree
/
Copy pathindex.html
More file actions
385 lines (358 loc) · 23.3 KB
/
index.html
File metadata and controls
385 lines (358 loc) · 23.3 KB
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
<!DOCTYPE html>
<html lang="en">
<head>
<meta charset="UTF-8">
<meta name="viewport" content="width=device-width, initial-scale=1.0">
<title>Marine Structures AI Tutor — Agentic RAG System</title>
<style>
:root {
--bg-deep: #060d1a;
--bg-card: #0a1628;
--bg-card-hover: #0d1d36;
--cyan: #00d4ff;
--cyan-dim: rgba(0,212,255,0.12);
--cyan-glow: rgba(0,212,255,0.25);
--gold: #ffb347;
--gold-dim: rgba(255,179,71,0.12);
--text: #e2e8f0;
--text-dim: #8899aa;
--text-muted: #556677;
--border: rgba(255,255,255,0.06);
--radius: 12px;
--radius-lg: 20px;
--font: -apple-system, BlinkMacSystemFont, "Segoe UI", "Noto Sans SC", sans-serif;
--mono: "SF Mono", "Fira Code", "Cascadia Code", monospace;
}
* { margin:0; padding:0; box-sizing:border-box; }
html { scroll-behavior:smooth; }
body { background:var(--bg-deep); color:var(--text); font-family:var(--font); line-height:1.6; overflow-x:hidden; }
/* ── Navigation ── */
nav { position:fixed; top:0; left:0; right:0; z-index:100; backdrop-filter:blur(20px) saturate(180%);
background:rgba(6,13,26,0.85); border-bottom:1px solid var(--border); padding:0 2rem; }
nav .nav-inner { max-width:1200px; margin:0 auto; display:flex; align-items:center; height:56px; justify-content:space-between; }
nav .logo { font-weight:700; font-size:1rem; color:var(--cyan); letter-spacing:-0.3px; }
nav .logo span { color:var(--text-dim); font-weight:400; margin-left:6px; }
nav a { color:var(--text-dim); text-decoration:none; font-size:0.85rem; margin-left:2rem; transition:color .2s; }
nav a:hover { color:var(--cyan); }
/* ── Hero ── */
.hero { min-height:100vh; display:flex; align-items:center; justify-content:center; text-align:center;
padding:6rem 2rem 4rem; position:relative; overflow:hidden; }
.hero::before { content:""; position:absolute; inset:0;
background:radial-gradient(ellipse 80% 60% at 50% 30%, rgba(0,212,255,0.06) 0%, transparent 60%),
radial-gradient(ellipse 60% 50% at 20% 80%, rgba(255,179,71,0.04) 0%, transparent 60%); }
.hero .hero-content { position:relative; z-index:1; max-width:900px; }
.hero .badge { display:inline-flex; align-items:center; gap:8px; background:var(--cyan-dim); border:1px solid rgba(0,212,255,0.15);
color:var(--cyan); padding:6px 18px; border-radius:100px; font-size:0.82rem; font-weight:500; margin-bottom:2rem; }
.hero .badge .dot { width:7px; height:7px; background:var(--cyan); border-radius:50%; animation:pulse 2s ease-in-out infinite; }
@keyframes pulse { 0%,100%{opacity:1} 50%{opacity:.3} }
.hero h1 { font-size:clamp(2.4rem,6vw,4.5rem); font-weight:800; line-height:1.1; letter-spacing:-1.5px; margin-bottom:1.5rem; }
.hero h1 .gradient { background:linear-gradient(135deg, var(--cyan) 0%, #7dd3fc 40%, var(--gold) 100%);
-webkit-background-clip:text; -webkit-text-fill-color:transparent; background-clip:text; }
.hero .subtitle { font-size:1.25rem; color:var(--text-dim); max-width:650px; margin:0 auto 2.5rem; line-height:1.7; }
.hero .cta-row { display:flex; gap:1rem; justify-content:center; flex-wrap:wrap; }
.btn { display:inline-flex; align-items:center; gap:8px; padding:12px 28px; border-radius:100px; font-size:0.95rem;
font-weight:600; cursor:pointer; transition:all .25s; text-decoration:none; border:none; }
.btn-primary { background:var(--cyan); color:var(--bg-deep); }
.btn-primary:hover { background:#33dfff; transform:translateY(-1px); box-shadow:0 8px 30px var(--cyan-glow); }
.btn-outline { background:transparent; border:1px solid rgba(255,255,255,0.15); color:var(--text); }
.btn-outline:hover { border-color:var(--cyan); color:var(--cyan); }
/* ── Section Common ── */
section { padding:6rem 2rem; }
.section-inner { max-width:1100px; margin:0 auto; }
.section-label { display:inline-block; font-family:var(--mono); font-size:0.75rem; color:var(--cyan); text-transform:uppercase;
letter-spacing:2px; margin-bottom:1rem; }
.section-title { font-size:clamp(1.8rem,3.5vw,2.8rem); font-weight:700; letter-spacing:-0.8px; margin-bottom:1rem; }
.section-desc { color:var(--text-dim); font-size:1.1rem; max-width:600px; margin-bottom:3rem; line-height:1.7; }
/* ── RAG Evolution ── */
.evo-grid { display:grid; grid-template-columns:repeat(4,1fr); gap:1px; background:var(--border); border-radius:var(--radius-lg); overflow:hidden; }
.evo-item { background:var(--bg-card); padding:2rem 1.5rem; text-align:center; position:relative; }
.evo-item .stage { font-family:var(--mono); font-size:0.7rem; color:var(--cyan); text-transform:uppercase; letter-spacing:1.5px; margin-bottom:0.75rem; }
.evo-item .arch { font-family:var(--mono); font-size:0.8rem; color:var(--text); margin-bottom:0.5rem; }
.evo-item .innov { font-size:0.8rem; color:var(--text-dim); line-height:1.5; }
.evo-item .arrow { display:none; }
@media(max-width:768px){ .evo-grid{grid-template-columns:1fr} .evo-item{padding:1.5rem} }
/* ── Architecture Viz ── */
.arch-viz { background:var(--bg-card); border-radius:var(--radius-lg); border:1px solid var(--border); overflow:hidden; }
.arch-row { display:flex; align-items:center; justify-content:center; gap:1rem; padding:1.5rem 2rem; flex-wrap:wrap; }
.arch-row + .arch-row { border-top:1px solid var(--border); }
.arch-label { font-family:var(--mono); font-size:0.7rem; color:var(--text-muted); text-transform:uppercase; letter-spacing:1.5px;
writing-mode:vertical-lr; margin-right:1rem; }
@media(max-width:768px){ .arch-row{gap:0.5rem;padding:1rem} }
.agent-node { background:var(--bg-card-hover); border:1px solid var(--border); border-radius:var(--radius);
padding:0.85rem 1.2rem; text-align:center; min-width:110px; transition:all .3s; }
.agent-node:hover { border-color:var(--cyan); box-shadow:0 0 20px var(--cyan-dim); transform:translateY(-2px); }
.agent-node .icon { font-size:1.3rem; margin-bottom:0.3rem; }
.agent-node .name { font-weight:600; font-size:0.85rem; }
.agent-node .role { font-size:0.7rem; color:var(--text-dim); margin-top:2px; }
.agent-node.supervisor { border-color:var(--gold); background:var(--gold-dim); }
.agent-node.supervisor .name { color:var(--gold); }
.arch-arrow { color:var(--text-muted); font-size:0.8rem; }
.arch-layer { background:rgba(0,212,255,0.03); border:1px dashed rgba(0,212,255,0.1); border-radius:var(--radius);
padding:1rem 2rem; text-align:center; }
.arch-layer .title { font-family:var(--mono); font-size:0.7rem; color:var(--cyan); text-transform:uppercase; letter-spacing:1.5px; margin-bottom:0.5rem; }
.arch-layer .items { font-size:0.82rem; color:var(--text-dim); }
/* ── Features Grid ── */
.features { display:grid; grid-template-columns:repeat(auto-fit,minmax(280px,1fr)); gap:1.5rem; }
.feature-card { background:var(--bg-card); border:1px solid var(--border); border-radius:var(--radius-lg);
padding:2rem; transition:all .3s; position:relative; overflow:hidden; }
.feature-card::before { content:""; position:absolute; top:0; left:0; right:0; height:2px;
background:linear-gradient(90deg, transparent, var(--cyan), transparent); opacity:0; transition:opacity .3s; }
.feature-card:hover { border-color:rgba(0,212,255,0.2); transform:translateY(-4px); box-shadow:0 20px 50px rgba(0,0,0,0.3); }
.feature-card:hover::before { opacity:1; }
.feature-card .f-icon { width:40px; height:40px; border-radius:10px; display:flex; align-items:center; justify-content:center;
font-size:1.2rem; margin-bottom:1.2rem; }
.feature-card .f-icon.cyan { background:var(--cyan-dim); color:var(--cyan); }
.feature-card .f-icon.gold { background:var(--gold-dim); color:var(--gold); }
.feature-card h3 { font-size:1.1rem; margin-bottom:0.6rem; letter-spacing:-0.3px; }
.feature-card p { font-size:0.88rem; color:var(--text-dim); line-height:1.65; }
/* ── Tech Stack ── */
.tech-grid { display:grid; grid-template-columns:repeat(auto-fit,minmax(160px,1fr)); gap:1rem; }
.tech-item { background:var(--bg-card); border:1px solid var(--border); border-radius:var(--radius);
padding:1.2rem; text-align:center; transition:all .25s; }
.tech-item:hover { border-color:var(--cyan); background:var(--bg-card-hover); }
.tech-item .t-name { font-weight:600; font-size:0.88rem; margin-bottom:0.3rem; }
.tech-item .t-why { font-size:0.73rem; color:var(--text-dim); line-height:1.5; }
/* ── Data Pipeline ── */
.data-table { width:100%; border-collapse:collapse; }
.data-table th { text-align:left; font-family:var(--mono); font-size:0.7rem; color:var(--text-muted); text-transform:uppercase;
letter-spacing:1.5px; padding:0.8rem 1rem; border-bottom:1px solid var(--border); }
.data-table td { padding:1rem; border-bottom:1px solid var(--border); font-size:0.88rem; }
.data-table tr:hover td { background:rgba(0,212,255,0.02); }
.data-table .num { font-family:var(--mono); color:var(--cyan); font-weight:600; }
/* ── Iteration Timeline ── */
.timeline { position:relative; padding-left:2rem; }
.timeline::before { content:""; position:absolute; left:8px; top:0; bottom:0; width:2px; background:var(--border); }
.tl-item { position:relative; margin-bottom:2.5rem; padding-left:2rem; }
.tl-item::before { content:""; position:absolute; left:-2rem; top:4px; width:18px; height:18px; border-radius:50%;
background:var(--bg-deep); border:2px solid var(--cyan); }
.tl-item.done::before { background:var(--cyan); border-color:var(--cyan); }
.tl-item .tl-num { font-family:var(--mono); font-size:0.7rem; color:var(--cyan); margin-bottom:0.3rem; }
.tl-item h4 { font-size:1.05rem; margin-bottom:0.3rem; }
.tl-item p { font-size:0.85rem; color:var(--text-dim); }
/* ── Footer ── */
footer { border-top:1px solid var(--border); padding:3rem 2rem; text-align:center; }
footer p { color:var(--text-muted); font-size:0.85rem; }
footer a { color:var(--cyan); text-decoration:none; }
/* ── Scroll Reveal ── */
.reveal { opacity:0; transform:translateY(30px); transition:opacity .8s, transform .8s; }
.reveal.visible { opacity:1; transform:translateY(0); }
/* ── Responsive ── */
@media(max-width:768px){
nav a { margin-left:1rem; font-size:0.78rem; }
.hero { padding:5rem 1.5rem 3rem; }
.features { grid-template-columns:1fr; }
.tech-grid { grid-template-columns:repeat(2,1fr); }
}
/* Floating particles */
.particles { position:fixed; inset:0; pointer-events:none; z-index:0; }
.particle { position:absolute; border-radius:50%; background:var(--cyan); opacity:0; animation:float 8s ease-in-out infinite; }
@keyframes float { 0%{transform:translateY(0) scale(1);opacity:0} 10%{opacity:.15}
90%{opacity:.15} 100%{transform:translateY(-400px) scale(.3);opacity:0} }
</style>
</head>
<body>
<nav>
<div class="nav-inner">
<div class="logo">Marine Structures AI Tutor<span>SESS3026/JEIS3005</span></div>
<div>
<a href="#architecture">Architecture</a>
<a href="#features">Features</a>
<a href="#tech">Tech Stack</a>
<a href="#iterations">Iterations</a>
<a href="https://github.com/ZSHYC/cer-rag" target="_blank">GitHub ↗</a>
</div>
</div>
</nav>
<section class="hero">
<div class="particles" id="particles"></div>
<div class="hero-content">
<div class="badge"><span class="dot"></span> Agentic RAG · Multi-Agent System</div>
<h1>An <span class="gradient">AI Exam Tutor</span><br>for Marine Structures</h1>
<p class="subtitle">A LangGraph-powered multi-agent system with Self-RAG reflection, three-layer indexing, and knowledge graph reasoning — demonstrating the full evolution from naive RAG to production-grade agentic RAG.</p>
<div class="cta-row">
<a href="https://github.com/ZSHYC/cer-rag" target="_blank" class="btn btn-primary">View on GitHub</a>
<a href="#architecture" class="btn btn-outline">Explore Architecture</a>
</div>
</div>
</section>
<section id="evolution">
<div class="section-inner">
<span class="section-label">01 · RAG Evolution</span>
<h2 class="section-title">From Naive RAG to Agentic RAG</h2>
<p class="section-desc">This project is not a "plug data into LangChain" demo. It traces the full RAG evolution path.</p>
<div class="evo-grid">
<div class="evo-item">
<div class="stage">Stage 1</div>
<div class="arch">Naive RAG</div>
<div class="innov">Dense + Sparse Hybrid<br>RRF Fusion</div>
</div>
<div class="evo-item">
<div class="stage">Stage 2</div>
<div class="arch">Advanced RAG</div>
<div class="innov">Cross-encoder Rerank<br>Multi-strategy Chunking</div>
</div>
<div class="evo-item">
<div class="stage">Stage 3</div>
<div class="arch"><strong>Agentic RAG</strong></div>
<div class="innov">Supervisor + 5 Specialist Agents<br>Self-RAG Reflection Loop</div>
</div>
<div class="evo-item">
<div class="stage">Stage 4</div>
<div class="arch">Production</div>
<div class="innov">FastAPI SSE + Docker<br>RAGAS Evaluation + A/B Testing</div>
</div>
</div>
</div>
</section>
<section id="architecture">
<div class="section-inner">
<span class="section-label">02 · System Architecture</span>
<h2 class="section-title">Multi-Agent Orchestration</h2>
<p class="section-desc">LangGraph state machine coordinates 6 specialized agents through a Supervisor pattern.</p>
<div class="arch-viz">
<div class="arch-row"><span class="arch-label">Interface</span>
<span class="arch-layer" style="flex:1"><span class="title">CLI (Rich) · Gradio UI · FastAPI SSE</span></span>
</div>
<div class="arch-row"><span class="arch-label">Agents</span>
<div class="agent-node supervisor"><div class="icon">🧠</div><div class="name">Supervisor</div><div class="role">Plan & Route</div></div>
<span class="arch-arrow">→</span>
<div class="agent-node"><div class="icon">🔍</div><div class="name">Retrieval</div><div class="role">Search + Self-RAG</div></div>
<span class="arch-arrow">→</span>
<div class="agent-node"><div class="icon">📖</div><div class="name">Concept</div><div class="role">Explain Theory</div></div>
<div class="agent-node"><div class="icon">📐</div><div class="name">Calc</div><div class="role">Solve Problems</div></div>
<div class="agent-node"><div class="icon">📝</div><div class="name">Exam</div><div class="role">Generate & Grade</div></div>
<div class="agent-node"><div class="icon">✅</div><div class="name">Critic</div><div class="role">Validate Output</div></div>
</div>
<div class="arch-row"><span class="arch-label">Tools</span>
<span class="arch-layer" style="flex:1"><span class="title">Shared Tool Layer</span>
<span class="items">dense_search · sparse_search · graph_traverse · formula_lookup · rerank</span></span>
</div>
<div class="arch-row"><span class="arch-label">Index</span>
<span class="arch-layer" style="flex:1"><span class="title">Three-Layer Indexing</span>
<span class="items">Dense Vectors (ChromaDB+BGE-M3) · Sparse Keywords (BM25) · Knowledge Graph (NetworkX)</span></span>
</div>
<div class="arch-row"><span class="arch-label">Infra</span>
<span class="arch-layer" style="flex:1"><span class="title">Infrastructure</span>
<span class="items">Docker Compose · LangSmith · RAGAS · CI/CD</span></span>
</div>
</div>
</div>
</section>
<section id="features">
<div class="section-inner">
<span class="section-label">03 · Key Features</span>
<h2 class="section-title">What Makes This Stand Out</h2>
<p class="section-desc">Every component is deliberately chosen and production-hardened.</p>
<div class="features">
<div class="feature-card reveal">
<div class="f-icon cyan">🤖</div>
<h3>6-Agent System</h3>
<p>Supervisor orchestrates Retrieval (with Self-RAG reflection), Concept, Calculation, Exam, and Critic agents — each specialized for its task.</p>
</div>
<div class="feature-card reveal">
<div class="f-icon gold">🔄</div>
<h3>Self-RAG Reflection</h3>
<p>Retrieval agent evaluates result sufficiency and rewrites queries for re-retrieval — inspired by Self-RAG (Asai et al., 2023) and Corrective RAG (Yan et al., 2024).</p>
</div>
<div class="feature-card reveal">
<div class="f-icon cyan">🗂️</div>
<h3>Three-Layer Indexing</h3>
<p>Dense vectors (BGE-M3) + Sparse keywords (BM25) + Knowledge Graph (NetworkX, 79 nodes, 269 edges) for structured concept-formula-exam traversal.</p>
</div>
<div class="feature-card reveal">
<div class="f-icon gold">🌐</div>
<h3>Cross-lingual</h3>
<p>BGE-M3 multilingual embeddings handle English lectures and Chinese study notes in one unified semantic space — no translation needed.</p>
</div>
<div class="feature-card reveal">
<div class="f-icon cyan">📊</div>
<h3>Evaluation-driven</h3>
<p>RAGAS metrics (faithfulness, context precision/recall, answer relevancy) with A/B strategy comparison — data-driven, not gut-feel.</p>
</div>
<div class="feature-card reveal">
<div class="f-icon gold">🚀</div>
<h3>Production-ready</h3>
<p>FastAPI + SSE streaming + Docker Compose one-command deployment + LangSmith observability — not a notebook, a deployable system.</p>
</div>
</div>
</div>
</section>
<section id="data">
<div class="section-inner">
<span class="section-label">04 · Data Pipeline</span>
<h2 class="section-title">738 Chunks Across 5 Sources</h2>
<p class="section-desc">Source-aware chunking strategy — the most critical hyperparameter for retrieval quality.</p>
<div style="overflow-x:auto; background:var(--bg-card); border:1px solid var(--border); border-radius:var(--radius-lg);">
<table class="data-table">
<thead><tr><th>Source</th><th>Documents</th><th>Chunks</th><th>Strategy</th><th>Metadata</th></tr></thead>
<tbody>
<tr><td>Lecture Slides</td><td class="num">21 PDFs · 582 pages</td><td class="num">200</td><td>3-slide merged windows</td><td>lecture_number, slide_range, topic_tags</td></tr>
<tr><td>Exam Papers</td><td class="num">9 years (2014–2025)</td><td class="num">25</td><td>Question-boundary split</td><td>year, question_number, marks, content_type</td></tr>
<tr><td>Answer Keys</td><td class="num">3 years (2022–2025)</td><td class="num">15</td><td>Sub-question paired chunks</td><td>year, question_number, sub_question</td></tr>
<tr><td>Seminar Solutions</td><td class="num">4 Tutorials</td><td class="num">23</td><td>Tutorial → Question hierarchical</td><td>tutorial_number, language</td></tr>
<tr><td>Chinese Study Notes</td><td class="num">12 Markdown files</td><td class="num">475</td><td>Markdown header-aware</td><td>file_topic, breadcrumb, language</td></tr>
<tr style="font-weight:700"><td>Total</td><td></td><td class="num">738</td><td colspan="2">Each chunk carries rich metadata for filtered/scoped retrieval</td></tr>
</tbody>
</table>
</div>
</div>
</section>
<section id="tech">
<div class="section-inner">
<span class="section-label">05 · Tech Stack</span>
<h2 class="section-title">Every Choice Has a Reason</h2>
<p class="section-desc">No defaults-for-defaults'-sake. Each component selected via deliberate trade-off analysis.</p>
<div class="tech-grid">
<div class="tech-item"><div class="t-name">DeepSeek V4 Pro</div><div class="t-why">LLM — Strong Chinese+math reasoning</div></div>
<div class="tech-item"><div class="t-name">LangGraph</div><div class="t-why">Agent Framework — State machine multi-agent orchestration</div></div>
<div class="tech-item"><div class="t-name">BGE-M3</div><div class="t-why">Embedding — 1024-dim, multilingual, local & free</div></div>
<div class="tech-item"><div class="t-name">ChromaDB</div><div class="t-why">Vector DB — Zero-ops, metadata filtering</div></div>
<div class="tech-item"><div class="t-name">BM25</div><div class="t-why">Sparse Retrieval — Exact formula/term matching</div></div>
<div class="tech-item"><div class="t-name">BGE-Reranker-v2</div><div class="t-why">Cross-encoder precision boost</div></div>
<div class="tech-item"><div class="t-name">NetworkX</div><div class="t-why">Knowledge Graph — 79 nodes, 269 edges</div></div>
<div class="tech-item"><div class="t-name">FastAPI + SSE</div><div class="t-why">Backend — Async streaming responses</div></div>
<div class="tech-item"><div class="t-name">Gradio + Rich</div><div class="t-why">UI — Web interface + polished CLI</div></div>
<div class="tech-item"><div class="t-name">RAGAS</div><div class="t-why">Evaluation — 5 ground-truth test cases</div></div>
<div class="tech-item"><div class="t-name">Docker Compose</div><div class="t-why">One-command production deployment</div></div>
<div class="tech-item"><div class="t-name">LangSmith</div><div class="t-why">Observability — Agent call chain tracing</div></div>
</div>
</div>
</section>
<section id="iterations">
<div class="section-inner">
<span class="section-label">06 · Development Journey</span>
<h2 class="section-title">5 Iterations, 10 Days</h2>
<p class="section-desc">Each iteration builds on the last, deliberately growing complexity.</p>
<div class="timeline">
<div class="tl-item done"><div class="tl-num">ITERATION 0</div><h4>Basic RAG</h4><p>Document processing, 738 chunks, three-layer indexing (dense+sparse+KG placeholder), hybrid retrieval with RRF fusion, CLI with Rich</p></div>
<div class="tl-item done"><div class="tl-num">ITERATION 1</div><h4>Agentic Upgrade</h4><p>LangGraph state graph, Supervisor agent with planning+routing, Retrieval agent with Self-RAG reflection loop, Concept & Calc specialist agents</p></div>
<div class="tl-item done"><div class="tl-num">ITERATION 2</div><h4>Knowledge Graph</h4><p>79-node, 269-edge knowledge graph (Concept↔Formula↔Lecture↔Exam), graph-traversal enhanced search, Exam Agent (generate + grade), Critic Agent (hallucination check)</p></div>
<div class="tl-item done"><div class="tl-num">ITERATION 3</div><h4>Productionization</h4><p>FastAPI backend with SSE streaming, Docker Compose deployment, RAGAS evaluation framework with 5 ground-truth cases, A/B strategy comparison</p></div>
<div class="tl-item done"><div class="tl-num">ITERATION 4</div><h4>Polish</h4><p>Gradio Web UI, comprehensive testing, bilingual README (Chinese+English), project website</p></div>
</div>
</div>
</section>
<footer>
<p>Built with LangGraph · DeepSeek V4 Pro · ChromaDB · BGE-M3 · NetworkX · FastAPI</p>
<p style="margin-top:0.5rem"><a href="https://github.com/ZSHYC/cer-rag" target="_blank">github.com/ZSHYC/cer-rag</a> · MIT License</p>
</footer>
<script>
// ── Scroll reveal ──
const observer = new IntersectionObserver((entries) => {
entries.forEach(e => { if(e.isIntersecting) e.target.classList.add('visible'); });
}, { threshold: 0.15 });
document.querySelectorAll('.reveal').forEach(el => observer.observe(el));
// ── Floating particles ──
const container = document.getElementById('particles');
if(container) {
for(let i=0; i<18; i++) {
const p = document.createElement('div');
p.className = 'particle';
const size = Math.random()*3+1;
p.style.cssText = `width:${size}px;height:${size}px;left:${Math.random()*100}%;top:${60+Math.random()*40}%;animation-delay:${Math.random()*8}s;animation-duration:${6+Math.random()*10}s`;
container.appendChild(p);
}
}
</script>
</body>
</html>