chopratejas
|
2ce26438a0
|
Integrate DynamicContentDetector into CacheAligner (Phase 1)
- Add DynamicContentDetector integration for comprehensive dynamic content
detection (20+ patterns vs previous 4 date patterns)
- New detection: UUIDs, API keys, JWT tokens, Unix timestamps, request/trace
IDs, hex hashes (MD5/SHA1/SHA256), version numbers, high-entropy strings
- Add CacheAlignerConfig options: use_dynamic_detector, detection_tiers,
extra_dynamic_labels, entropy_threshold
- Maintain backward compatibility with legacy date-only mode
- Add 25 new comprehensive tests for Phase 1 functionality
- Fix code compressor fallback test to properly mock LLMLingua availability
Expected cache hit improvement: 30-50% by extracting more dynamic content
|
2026-01-19 22:56:20 -08:00 |
|
chopratejas
|
9c7d4512d6
|
Initial commit: Headroom SDK - LLM context optimization toolkit
A comprehensive SDK for optimizing LLM context windows, reducing token
usage while preserving critical information for AI agents.
Core Features:
- SmartCrusher: Statistical compression of tool outputs (70-85% reduction)
- CacheAligner: Prefix optimization for prompt cache hits
- RollingWindow: Intelligent context window management
- BM25/Hybrid relevance scoring for smart item selection
Integrations:
- OpenAI and Anthropic provider support
- LangChain integration (ChatModel, Callbacks, Runnable)
- MCP (Model Context Protocol) integration for tool compression
Test Coverage:
- 372 tests passing across all modules
- 35 performance benchmarks
- Real-world agent evaluations with 88% token savings
Key Components:
- headroom/transforms/: Core compression transforms
- headroom/providers/: OpenAI and Anthropic support
- headroom/integrations/: LangChain and MCP integrations
- headroom/relevance/: BM25 and hybrid scoring
- headroom/pricing/: Model pricing registry
- benchmarks/: Performance benchmark suite
- examples/: Usage examples and demos
|
2026-01-06 23:16:58 -08:00 |
|