/rag-langchain-demo

A RAG app with Open ai as LLM to chat with a list of documents.

Primary LanguagePythonOtherNOASSERTION

Running the Retrieval-Augmented Chat Application

1. Environment Setup

This project uses Poetry for dependency management.

  • If you don't have Poetry installed, follow the instructions on the official website: https://python-poetry.org/docs/#installation

  • Navigate to the project directory and run:

    poetry install

  • Create a .env file in the project directory and add:

    • PINECONE_API_KEY: Your Pinecone API key
    • PINECONE_ENV: Your Pinecone environment name (e.g., "dev")
    • OPENAI_API_KEY: Your OpenAI API key

  • Install dependencies: pip install streamlit pinecone-client langchain PyPDF2

2. Creating Embeddings

  • Run create_embeddings.py to embed PDF documents in files/ and store them in a Pinecone index: 
    python create_embeddings.py

3. Running the Chat App

  • Run chat.py to start the Streamlit chat interface: 
    streamlit run chat.py

4. Using the Chat App

  • Enter your username in the sidebar.
  • Ask questions or make statements in the chat box.
  • The app retrieves relevant information and generates responses using a large language model.
  • Clear the chat history with the sidebar button.

5. SQL History and Database

  • Chat history is stored in local_sqlite_db.db for persistence and context.
  • This allows for more personalized and consistent conversations.

Additional Notes

  • Ensure PDFs exist in files/ before running embeddings.py.
  • The app uses gpt-3.5-turbo from OpenAI, requiring an OpenAI API key.
  • Adjust embeddings model and retrieval parameters for fine-tuning.

License

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