pRAG

pRAG is an production-ready fully local question-answering system specifically designed scientific documentation. It leverages state-of-the-art retrieval-augmented generation (RAG) techniques to help students, researchers, and academics efficiently extract and synthesize information from academic papers and documents.

Features

• Document Processing: Upload and process multiple PDF files simultaneously with RAY
• Advanced RAG Pipeline:
  • Vector-based document retrieval using Qdrant
  • Context-aware response generation
  • PDF content display and reference
• User Authentication: Secure admin access with password protection
• VectorDB: Using Qdrant with docker to avoid InMemory
• Chat Interface: Interactive chat-based interface powered by Chainlit
• Multi-Model Support: Integration with various LLM providers (DeepSeek, OpenAI, Google)
• Docker Support: Containerized deployment with Docker and docker-compose

Project Structure

• Core Components:
  • app.py: Main application logic and Chainlit interface
  • db.py: Vector database management and document processing
  • config.py: Configuration management and API keys
  • chain.py: RAG pipeline implementation

Setup

1. Prerequisites:

   • Docker and Docker Compose
   • Python 3.8+
   • NVIDIA GPU

2. Environment Setup:

   # Clone the repository
   git clone https://github.com/KSonPham/pRAG.git
   cd pRAG
   
   # Create .env file with required environment variables
   cp .env.example .env
   # Create chainlit-secret:
   chainlit create-secret
   # Edit .env with your chainlit security key

3. Docker Deployment:

   docker-compose up -d

4. Local Development:

   pip install -r requirements.txt
   python app.py

Usage

1. Access the application through your web browser
2. Log in with admin/admin credentials
3. Upload PDF documents for processing
4. Ask questions about the uploaded documents

Security Notes

• The application uses bcrypt for password hashing
• API keys are stored in the configuration file
• Admin access is protected by password authentication

Advanced Features

• Multi-Document Analysis: Cross-document question answering
• Context-Aware Responses: Intelligent context retrieval and synthesis
• PDF Display: Interactive PDF viewing with relevant sections highlighted

Contributing

Contributions are welcome! Please feel free to submit a Pull Request.