Skip to content

Releases: Flamehaven/CRoM-Context-Rot-Mitigation--EfficientLLM

CRoM-EfficientLLM v1.0.2

17 Sep 12:28
@Flamehaven Flamehaven
v1.0.2
1b5b98e
Compare
Choose a tag to compare
CRoM-EfficientLLM v1.0.2 Latest
Latest

[1.0.2] - 2025-09-17

🐞 Fixed

  • Critical SyntaxError in SafeCrossEncoderManager.get_status_for_response
    • Separated docstring and return statement that were incorrectly on the same line
    • Resolved import failures preventing crom_efficientllm.cross_encoder module loading
  • FastAPI Server ImportError for enhanced_greedy_pack
    • Implemented lazy loading with importlib to fix circular import dependency
    • Corrected module structure so enhanced_greedy_pack is properly exported from budget_packer
    • Resolved FastAPI server startup failures
  • Version consistency issues across components
    • FastAPI server now correctly displays v1.0.2
    • Fixed dynamic version loading from pyproject.toml and package metadata
    • Synchronized version references across the codebase

⚡ Improved

  • More robust error handling and import resilience across core modules
  • Refined module architecture preventing circular dependencies
  • Clearer debugging output for Cross-Encoder initialization
  • Dynamic version management system with proper fallback mechanisms

✨ Added

  • Comprehensive integration tests (tests/test_integration.py)
    • End-to-end validation with 10 test cases
    • Verified version consistency across all components
    • Interoperability tests (CrossEncoder + BudgetPacker + Logger)
    • Performance testing with 1,000+ documents
    • Edge-case error handling and fallback validation
    • FastAPI server startup and endpoint checks

🧪 Testing

  • All core module imports function correctly
  • FastAPI server launches without errors and shows correct version
  • Cross-Encoder manager handles errors gracefully
  • Integration tests validate full system functionality
  • Performance benchmarks confirm scalability with large datasets
Assets 2
Loading

CRoM v1.0.1 – Context Rot Mitigation Core Release

06 Sep 05:44
@Flamehaven Flamehaven
v1.0.1
8f95629
Compare
Choose a tag to compare
CRoM v1.0.1 – Context Rot Mitigation Core Release

🧠 CRoM v1.0.1 – Context-Rot Mitigation for Efficient LLMs

Release Date: 2025-09-06
Tag: v1.0.1

🚀 What's New

First functional release of the CRoM system — a context efficiency layer for Retrieval-Augmented LLM systems. This version brings token-aware injection, cross-encoder fallback, and full FastAPI orchestration.

🔧 Features

  • budget_packer.py – Smart prompt packing under token budgets
  • Cross-Encoder Fallback – Embedding-based scoring when LLMs are uncertain
  • Capsule Logger – Transparent logging of injected/pruned context
  • FastAPI Interface – Live endpoint access for simulation, injection, and logging

⚠️ Known Issues

  • Persistent vector cache (FAISS) not yet implemented
  • Evaluation tests are internal only
  • Single-thread fallback slows batch performance

🔭 Roadmap

  • Vector memory cache
  • Context summarizer & history pruning
  • Performance benchmarking suite

🔌 Install & Run

git clone https://github.com/Flamehaven/CRoM-Context-Rot-Mitigation--EfficientLLM.git
cd CRoM-Context-Rot-Mitigation--EfficientLLM
pip install -r requirements.txt
uvicorn app.main:app --reload
Loading