LLM-Powered-Document-Q-A-with-Llama-3.2

A Retrieval-Augmented Generation (RAG) application that enables intelligent document analysis and question answering using Llama 3.2. Built with Streamlit, Langchain, and Ollama.

Features

• Multi-format Document Support: Process PDF, DOCX, TXT, MD, and CSV files with intelligent text extraction.
• Advanced Text Processing: Automatic text chunking and embedding using LangChain.
• Vector Search: FAISS-powered similarity search for accurate document retrieval.
• Configurable Parameters: Adjustable model settings, chunk sizes, and temperature.
• Comprehensive Metadata Tracking: Detailed processing statistics and response metrics.
• Interactive UI: Clean Streamlit interface with chat history and document statistics.

Prerequisites

• Python 3.8+
• Ollama installed and running locally
• Required Python packages:
  • ollama
  • streamlit
  • chardet
  • PyPDF2
  • python-docx
  • langchain
  • faiss-cpu

Installation

1. Clone the repository:

   git clone <repository-url>
   cd LLM-Powered-Document-Q-A-with-Llama-3.2

2. Install dependencies:

   pip install -r requirements.txt

3. Ensure Ollama is running locally on port 11434.

Usage

🚀 Quick Start

Install Ollama and pull Llama 3.2:

ollama pull llama3.2

Install dependencies:

pip install -r requirements.txt

Run the app:

streamlit run app.py

1. Start the application:

   streamlit run app.py

2. Access the web interface at http://localhost:8501.
3. Configure the application using the sidebar:
   • Select the LLM model.
   • Adjust temperature and chunk size.
   • Upload documents or paste text directly.
4. Enter questions and receive AI-powered responses based on your documents.

Architecture

Key Components

• DocumentProcessor

  • Handles multiple document formats.
  • Extracts text with metadata.
  • Provides error handling and encoding detection.

• EnhancedRAGApplication

  • Manages document indexing and retrieval.
  • Handles embedding generation.
  • Coordinates response generation.
  • Tracks performance metrics.

• Streamlit UI

  • Provides interactive interface.
  • Displays document statistics.
  • Maintains chat history.
  • Shows response metadata.

Configuration Options

• model_name: LLM model selection (default: "llama3.2")
• embedding_model: Model for generating embeddings
• chunk_size: Text chunk size (default: 500)
• chunk_overlap: Overlap between chunks (default: 50)
• temperature: Response creativity (0.0-1.0)
• max_tokens: Maximum response length

Error Handling

The application includes comprehensive error handling:

• Document processing validation
• Encoding detection and fallback
• Vector store initialization checks
• Response generation monitoring

Logging

Detailed logging is implemented throughout:

• Document processing events
• Vector store operations
• Response generation metrics
• Error tracking

Performance Considerations

• Document chunking is optimized for balance between context and performance.
• Vector similarity search uses efficient FAISS indexing.
• Response generation includes context length management.
• UI updates are optimized for responsiveness.

Contributing

1. Fork the repository.
2. Create a feature branch.
3. Implement changes with tests.
4. Submit a pull request.

License

MIT License

Acknowledgments

• Ollama team for the LLM integration.
• Streamlit for the UI framework.
• LangChain for document processing utilities.
• FAISS for vector search capabilities.