RAG - Retrieval Augmented Generation

• Tutorial Desktop: Video

• Tutorial Android: Video

Overview

RAG leverages Meta Llama 3 (8B parameters) on GPU and Hugging Face API models on CPU. It supports two primary functionalities:

1. Chat with LLM: Engage in conversations with the large language model using GPU.
2. RAG Chat with PDFs: Perform Retrieval-Augmented Generation with up to 4 PDF documents.

Features

• Chat with LLM: Utilize Meta Llama 3 for chat interactions, supporting system and user messages.
• RAG Chat with PDFs: Interact with content from PDFs using various prompts:
  • Detailed Prompt
  • Short Prompt
  • Summary Prompt
  • Explanation Prompt
  • Opinion Prompt
  • Instruction Prompt

Setup and Installation

1. Clone the Repository:

   git clone https://github.com/SirajuddinShaik/RAG.git
   cd RAG

2. Install Dependencies:

   pip install -r requirements.txt

3. Run the Application:

   chainlit app.py

Usage

To run the application with GPU support, you can use the following Colab link: Run on Colab

Example Commands

• Start a chat session:

  chainlit app.py

• Load and interact with PDFs: Ensure your PDFs are in the appropriate directory and use the UI to upload and query them.

Project Structure

• app.py: Main application script.
• requirements.txt: Dependencies required for the project.
• src/: Source code directory.
• config/: Configuration files.
• src/utils/prompts: various prompts used to interact.
• data_ingestion/: Scripts and tools for data ingestion.
• logs/: Log files.
• Dockerfile: Docker setup for containerized deployment.

Contributing

We welcome contributions! Please fork the repository, create a new branch, and submit a pull request.

License

This project is licensed under the MIT License.

---

If you have any questions or need further assistance, please open an issue on the GitHub repository.

---

Happy Coding! 🚀