File size: 2,709 Bytes
88f4fd7 e4d5155 |
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 |
---
library_name: efficient-context
language: code
tags:
- context-optimization
- llm
- cpu-optimization
- resource-constrained
- memory-management
- python
license: mit
datasets:
- None
---
<!-- filepath: /Users/biswanath2.roul/Desktop/biswanath/office/poc/pypi/190525/efficient-context/README.md -->
# efficient-context
A Python library for optimizing LLM context handling in CPU-constrained environments.
## Overview
`efficient-context` addresses the challenge of working with large language models (LLMs) on CPU-only and memory-limited systems by providing efficient context management strategies. The library focuses on:
- **Context Compression**: Reduce memory requirements while preserving information quality
- **Semantic Chunking**: Go beyond token-based approaches for more effective context management
- **Retrieval Optimization**: Minimize context size through intelligent retrieval strategies
- **Memory Management**: Handle large contexts on limited hardware resources
## Installation
```bash
pip install efficient-context
```
## Quick Start
```python
from efficient_context import ContextManager
from efficient_context.compression import SemanticDeduplicator
from efficient_context.chunking import SemanticChunker
from efficient_context.retrieval import CPUOptimizedRetriever
# Initialize a context manager with custom strategies
context_manager = ContextManager(
compressor=SemanticDeduplicator(threshold=0.85),
chunker=SemanticChunker(chunk_size=256),
retriever=CPUOptimizedRetriever(embedding_model="lightweight")
)
# Add documents to your context
context_manager.add_documents(documents)
# Generate optimized context for a query
optimized_context = context_manager.generate_context(query="Tell me about the climate impact of renewable energy")
# Use the optimized context with your LLM
response = your_llm_model.generate(prompt=prompt, context=optimized_context)
```
## Features
### Context Compression
- Semantic deduplication to remove redundant information
- Importance-based pruning that keeps critical information
- Automatic summarization of less relevant sections
### Advanced Chunking
- Semantic chunking that preserves logical units
- Adaptive chunk sizing based on content complexity
- Chunk relationships mapping for coherent retrieval
### Retrieval Optimization
- Lightweight embedding models optimized for CPU
- Tiered retrieval strategies (local vs. remote)
- Query-aware context assembly
### Memory Management
- Progressive loading/unloading of context
- Streaming context processing
- Memory-aware caching strategies
## Maintainer
This project is maintained by [Biswanath Roul](https://github.com/biswanathroul)
## License
MIT |