Advanced RAG System with Llama2 & LlamaIndex

The Advanced Retrieval Augmented Generation (RAG) System leverages Open Source Llama2 and LlamaIndex, along with HuggingFace, to offer a powerful solution for querying large collections of custom data documents. This README provides comprehensive setup and usage instructions to help you harness the full potential of this system.

Here's the Google Colab Notebook: https://colab.research.google.com/drive/1aOSyEDygYQSMbHxM92Gd4ZdgS7XKIb_i?usp=sharing

The data is a synthetic generated dataset containing information about different personality scores and collection of user queries and corresponding assistant responses.

Table of Contents

• Introduction
• Prerequisites
• Installation
• Getting Started
• Usage
• Features
• Contributing
• License

Prerequisites

Before setting up the RAG system, ensure you have:

• Python 3.10
• llama_index
• torch
• pypdf
• transformers
• einops
• accelerate
• langchain
• bitsandbytes
• sentence_transformers
• huggingface_hub
• A T4 GPU or better (for optimal performance)
• An active HuggingFace token for API access

Installation

To install and set up the RAG system:

1. Clone the repository:

   git clone https://github.com/your-username/rag-system-repo.git
   cd rag-system-repo
   

2. Install the required Python packages:

   pip install -r requirements.txt
   

3. Log in to your HuggingFace account to access models:

   huggingface-cli login
   

   Follow the prompts to enter your API key.

Getting Started

To initialize the system:

1. Prepare your documents (.pdf, .txt, .csv, .html etc) in a directory named data.

2. Run the main Python script to start indexing (incase running in local PC) or run the .ipynb notebook:

   python rag_llm_llama2_llamaindex.py
   

Usage

Query your indexed documents using natural language:

• Input your questions or search terms in the provided query interface.
• The system retrieves relevant information from your document collection based on the queries.

Features

• Efficient Document Indexing: Uses LlamaIndex for fast, scalable indexing of large document sets.
• Powerful Query Capabilities: Employs the Llama2 model from Meta for understanding and processing natural language queries.
• Flexible Integration: Designed for seamless operation with HuggingFace's transformers and other ML libraries.
• GPU Optimization: Optimized for T4 GPUs and above for high-speed processing and retrieval.

Contributing

To contribute to the RAG system:

1. Fork the repository.
2. Create a new branch for your feature (git checkout -b feature/AmazingFeature).
3. Commit your changes (git commit -m 'Add some AmazingFeature').
4. Push to the branch (git push origin feature/AmazingFeature).
5. Open a pull request.

License

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