RAG-Agent

A powerful Retrieval Augmented Generation (RAG) system that enables you to chat with your documents using local LLMs through Ollama.

Features

• 📄 PDF Processing: Upload and process PDF documents
• 💬 Interactive Chat: Streaming responses for natural conversation
• 🧠 Contextual Retrieval: RAG system pulls relevant information from your documents
• 📊 Database Management: Create, manage and search document collections
• 👤 User Authentication: Secure login system for multi-user support
• 📱 Responsive Design: Works well on desktop and mobile

Prerequisites

• Python 3.8+
• Ollama installed and running locally (https://ollama.ai)
• Models pulled into Ollama (e.g., ollama pull llama3.2:latest)

Installation

1. Clone this repository:

git clone https://github.com/yourusername/RAG-Agent.git
cd RAG-Agent

2. Install dependencies:

pip install -r requirements_streamlit.txt

Usage

1. Start the Streamlit app:

streamlit run app_streamlit.py

2. Open your browser and navigate to: http://localhost:8501

3. Working with the Application:

   • Login/Register: Create an account or login to access your document collections
   • Upload Documents: Use the file uploader to add PDF documents to your collection
   • Create Collections: Organize your documents into collections with descriptions
   • Chat with Documents: Ask questions and get answers based on your documents
   • Toggle Features: Enable/disable RAG to switch between document-based and general responses
   • Change Models: Select from available Ollama models through the dropdown

How It Works

The RAG-Agent system:

1. Processes uploaded PDF documents, breaking them into chunks
2. Creates vector embeddings of these chunks using sentence-transformers
3. Stores these embeddings in a ChromaDB vector database
4. On receiving a question, finds the most relevant document chunks
5. Sends the relevant context along with your question to the LLM
6. Returns a response grounded in your documents

Project Structure

• app_streamlit.py: Main Streamlit application
• rag-agent/src/: Core RAG system components
  • fileloader.py: Document processing and embedding
  • ollama_pure_chat.py: Interface with Ollama API
  • document_parser.py: Document parsing utilities

Configuration

Database and storage paths are configured to use the localdatabase/ directory by default. This can be changed by setting the PERSISTENT_STORAGE_BASE environment variable.

Troubleshooting

• Ollama Connection Issues: Ensure Ollama is running with ollama serve
• Model Not Found: Verify you've pulled the model with ollama pull modelname
• PDF Processing Errors: Try different PDF files or check if your PDF is searchable/OCR'd
• Database Issues: Check permissions on the localdatabase directory

Contributing

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

License

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