A Retrieval-Augmented Generation (RAG) chatbot application that leverages Groq's high-performance language models and LangChain framework to provide intelligent responses based on document content.

This project implements a document-based question answering system that:

• Loads and processes PDF documents from a specified directory
• Creates vector embeddings for semantic search
• Utilizes Groq's LLM for generating responses
• Maintains conversation history for contextual interactions
• Provides a clean Streamlit web interface

• Document Processing: Automatic loading and chunking of PDF documents
• Vector Store: ChromaDB integration for efficient document retrieval
• LLM Integration: Groq API connection with configurable models
• Conversation Memory: Maintains chat history throughout sessions
• User-Friendly Interface: Streamlit-based web UI with sidebar controls
• Logging System: Comprehensive logging with file rotation and error tracking

1. Clone the repository:

git clone git@github.com:zolppy/omnidoc.git
cd omnidoc

2. Install required dependencies:

pip install -r requirements.txt

3. Set up environment variables:

export GROQ_API_KEY=your_groq_api_key_here

omnidoc/
├── app.py                 # Main Streamlit application
├── README.md              # Project documentation (this file)
├── LICENSE                # MIT License file
├── requirements.txt       # Python dependencies
├── utils/
│   ├── __init__.py       # Package initialization
│   ├── logger.py         # Logging configuration and setup
│   ├── rag.py            # Document processing and vector store management
│   └── llm.py            # LLM model initialization
├── data/                 # Directory for PDF documents (create this)
├── logs/                 # Auto-generated log files directory
└── vector_store/         # Auto-generated vector store directory

• GROQ_API_KEY: Your Groq API key (required)

• Model Selection: Change the default model in utils/llm.py (default: "llama3-70b-8192")
• Document Directory: Modify the path in utils/rag.py (default: "data/")
• Chunk Settings: Adjust chunk size and overlap in split_documents()
• Logging: Configure log levels and file paths in utils/logger.py

1. Place your PDF documents in the data/ directory
2. Run the application:

streamlit run app.py

3. Open your browser to the provided local URL (typically http://localhost:8501)
4. Click "Initialize / Reload Model" in the sidebar to process documents
5. Start chatting with your documents!

• Initialize/Reload Model: Processes documents and sets up the vector store
• Restart Chat: Clears conversation history while keeping the model loaded

• Configurable logging to both console and files
• Rotating file handlers for main logs and error logs
• Custom formatting with timestamps, log levels, and source information

• Document loading from PDF directories
• Text splitting with configurable chunk parameters
• Vector store management with persistent storage
• Support for both building new and loading existing vector stores

• Groq API integration
• Configurable model selection and temperature
• Error handling for model initialization

• Streamlit-based web interface
• Conversation history management
• Retrieval chain implementation with context-aware responses

The application generates logs in the logs/ directory:

• app.log: General application logs with DEBUG level
• errors.log: Error-specific logs with rotating daily retention

1. GROQ_API_KEY not set: Ensure you've set the environment variable
2. No documents found: Place PDF files in the data/ directory
3. Vector store errors: Use "force_rebuild=True" or delete the vector_store directory to rebuild
4. Module import errors: Ensure you're running from the project root directory

Enable debug logging by modifying the console_level parameter in utils/logger.py:

console_level=logging.DEBUG

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

Contributions are welcome! Please feel free to submit pull requests or open issues for bugs and feature requests.

For questions or issues, please check the logs in the logs/ directory or open an issue in the project repository.