quick-rag

A quick and simple RAG implementation, reads from markdown and text data sources and provides a conversational chat bot.

TODO:

• Add memory, so the chatbot can remember details from the conversation.
• Wrap the ChatBot into a class that supports multiple concurrent conversations.
• Implement a client / server architecture to allow multiple concurrent conversations.
• Add a UI to simplify the interaction with the users.

Instructions

1. Place your OpenAI and LangChain API keys in RAGConfig.py

    __EnvironmentVariablesDict = {
        "LANGCHAIN_TRACING_V2": "true",
        "LANGCHAIN_API_KEY": "<PLACE YOUR LANGCHAIN API KEY HERE>",
        "OPENAI_API_KEY": "<PLACE YOUR OPENAI API KEY HERE>",
    }

2. Run python CreateDataBase.py to extract the info in data/books into a Chroma database, this will be the base of knowledge for our RAG. Be aware that you can add more markdown and text files to the books folder to expand the knowledge base.

3. Run python ChatBot.py and chat with the ChatBot interactively, remember that it is constrained to try to respond queries from the knowledge base. Try playing with it, asking questions in and out of the knowledge base to see how it responds.