Skip to content

Latest commit

 

History

History
66 lines (48 loc) · 2.07 KB

File metadata and controls

66 lines (48 loc) · 2.07 KB

GLYPH Research & Benchmark Reports

These reports are dated research snapshots.

They are useful for historical context, benchmarking notes, and earlier experiments, but they are not authoritative API or installation docs. Package names, examples, and positioning may have drifted since some of these reports were written.


Reports Index

Performance & Benchmarks

CODEC_BENCHMARK_REPORT.md

  • Codec performance (parsing, canonicalization, fingerprinting)
  • Speed metrics across languages (Go, Python, JS)
  • Token reduction measurements
  • Key finding: Go achieves 2M+ ops/sec canonicalization

OPTIMIZATION_REPORT.md

  • Performance optimization techniques
  • Memory usage analysis
  • Codec tuning results

BENCHMARK_INDEX.md

  • Index of all benchmark runs
  • Historical performance data
  • Comparison across versions

LLM Integration

LLM_ACCURACY_REPORT.md

  • How LLMs handle GLYPH vs JSON
  • Retrieval accuracy across formats
  • Generation quality comparison
  • Embedding analysis
  • Key finding: Hybrid approach recommended (LLMs generate JSON, store as GLYPH)

STREAMING_VALIDATION_REPORT.md

  • Streaming validation performance
  • Token-by-token error detection
  • Latency savings measurements
  • Key finding: Detect errors as they stream and cancel immediately, saving tokens and time

TOOL_CALL_REPORT.md

  • Tool calling patterns
  • System prompt token reduction
  • Validation accuracy
  • Error recovery strategies

Quick Links

Documentation:

Code:


Questions? Open an issue or check discussions.