Multi-Document RAG Assistant

This project is an AI powered document question answering application built with Python, LangChain, ChromaDB, Hugging Face, Groq, and Streamlit using Retrieval-Augmented Generation (RAG).

Overview

This application uses a Retrieval-Augmented Generation (RAG) pipeline to retrieve relevant information from uploaded documents and provide grounded answers using a Large Language Model (LLM).

Instead of relying only on the LLM's pretrained knowledge, the system retrieves relevant document content and provides it as context to the model before generating an answer. This helps make responses more relevant to the uploaded documents and reduces unsupported answers.

Video Demo

Watch the Multi-Document RAG Assistant Demo

Features

• Ask natural language questions about uploaded documents.
• Retrieves relevant document content before generating an answer.
• Uses Hugging Face sentence transformer embeddings to represent document content as vectors.
• Uses ChromaDB for storing and retrieving document embeddings.
• The LLM is instructed to answer using the retrieved document context rather than relying solely on general knowledge.
• Retrieved document sections can be displayed alongside generated answers.
• Provides a simple web interface for uploading documents and asking questions.

Technologies Used

• Python
• LangChain
• ChromaDB
• Hugging Face
• Groq
• Streamlit

Basic Requirements

• Python 3.10+
• Groq API key
• Hugging Face API key

Python dependencies are listed in Requirements.txt

Installation

1. Clone the repository

   • git clone https://github.com/CS-Alfred/Multi-Document-RAG-Assistant.git
     1. Navigate into project folder
        • cd Multi-Document-RAG-Assistant

2. Create a virtual environment

   • python -m venv venv
     1. To activate in windows
        • venv\Scripts\activate
     2. To activate in macOS/Linux
        • source venv/bin/activate

3. Install Dependencies

   • pip install -r Requirements.txt

4. Configure API Keys

   • Create a .env file in the project root HUGGINGFACEHUB_API_KEY=your_huggingface_api_key GROQ_API_KEY=your_groq_api_key

   Replace the placeholder values with your own API keys.

   Important: Never commit your .env file or expose your API keys publicly.

5. Run the Application

   • streamlit run app.py

   The application will open in your browser.

References and Resources

This project was developed by studying and referring to documentation, tutorials, videos, and resources related to RAG, LangChain, vector databases, embeddings, and LLM applications.

Documentation

• LangChain Documentation
• ChromaDB Documentation
• Hugging Face Documentation
• Groq Documentation
• Streamlit Documentation

YouTube

• Large Language Models explained briefly
• Transformers, the tech behind LLMs
• Attention in transformers, step-by-step

Author

C S Alfred