CAG Demonstrator Agent

This project implements a demonstrator agent that compares the Cache-Augmented Generation (CAG) Framework with traditional Retrieval-Augmented Generation (RAG) using various LLMs.

Features

• Implements both CAG and RAG frameworks for comparison
• Supports multiple LLM providers (OpenAI, Anthropic, Google, Mistral, Groq)
• Measures and compares performance metrics
• Generates detailed comparison reports
• Uses efficient caching mechanisms for improved performance

Installation

1. Clone the repository:

git clone https://github.com/ai-in-pm/Cache-Augmented-Generation-CAG.git
cd Cache-Augmented-Generation-CAG

2. Create a virtual environment and activate it:

python -m venv venv
# On Windows
venv\Scripts\activate
# On Unix or MacOS
source venv/bin/activate

3. Install dependencies:

pip install -r requirements.txt

4. Set up environment variables:

• Copy .env.example to .env
• Add your API keys for the LLM providers you want to use:

  OPENAI_API_KEY=your_key_here
  ANTHROPIC_API_KEY=your_key_here
  MISTRAL_API_KEY=your_key_here
  GROQ_API_KEY=your_key_here
  GOOGLE_API_KEY=your_key_here
  

Usage

Run the demonstrator:

python demonstrator.py

The demonstrator will:

1. Initialize both CAG and RAG frameworks
2. Run a series of comparison queries
3. Generate metrics and save results to the Results directory

Project Structure

CAG/
├── cag_demo/                  # Main package directory
│   ├── __init__.py
│   ├── cag_framework.py      # CAG implementation
│   ├── rag_framework.py      # RAG implementation
│   ├── llm_interface.py      # LLM API interface
│   └── config.py             # Configuration settings
├── Data/                     # Data directory
│   ├── Preloaded_Contexts/   # CAG knowledge base
│   └── Retrieved_Documents/  # RAG document store
├── Results/                  # Comparison results
├── demonstrator.py           # Main demonstration script
├── requirements.txt          # Project dependencies
└── .env.example             # Environment variables template

Framework Comparison

The demonstrator compares two approaches:

1. Cache-Augmented Generation (CAG):

   • Preloads and caches knowledge
   • Eliminates real-time retrieval steps
   • Reduces latency and improves response times
   • Uses efficient memory management

2. Retrieval-Augmented Generation (RAG):

   • Traditional document retrieval approach
   • Real-time document fetching
   • Standard context processing

Results

Results are saved in JSON format in the Results directory with the following information:

• Timestamp
• LLM model used
• Framework configurations
• Query responses
• Performance metrics
• Time comparisons

Contributing

1. Fork the repository
2. Create your feature branch (git checkout -b feature/amazing-feature)
3. Commit your changes (git commit -m 'Add amazing feature')
4. Push to the branch (git push origin feature/amazing-feature)
5. Open a Pull Request

License

This project is licensed under the MIT License - see the LICENSE file for details.

Acknowledgments

• Thanks to all contributors and the open-source community
• Inspired by advances in LLM architectures and retrieval techniques