RAG Chatbot

A Retrieval-Augmented Generation (RAG) chatbot that uses Ollama's phi4 model for intelligent question answering over documents.

Features

• PDF document processing and text extraction
• Intelligent text chunking and embedding generation
• Vector-based semantic search
• RAG-based question answering
• LangGraph workflow for state management
• Configurable model parameters
• Progress tracking and error handling
• Modern Streamlit UI

System Architecture

Data Flow

The system follows a RAG architecture with the following components:

1. Document Processing

   • PDF text extraction
   • Text chunking
   • Embedding generation
   • Vector store creation

2. Query Processing

   • User question input
   • Context retrieval
   • Response generation
   • Answer presentation

For a detailed view of the data flow, see the RAG Flow Diagram.

LangGraph Workflow

The system uses LangGraph for state management and workflow control:

1. State Management

   • AgentState for tracking question, context, and answer
   • TypedDict for type-safe state handling
   • Clear state transitions

2. Processing Nodes

   • Retrieve node for context gathering
   • Generate node for answer creation
   • Error handling and logging

For a detailed view of the workflow, see the LangGraph Workflow Diagram.

Installation

1. Clone the repository:

git clone https://github.com/yourusername/rag-chatbot.git
cd rag-chatbot

2. Create and activate a virtual environment:

python -m venv venv
source venv/bin/activate  # On Windows: venv\Scripts\activate

3. Install dependencies:

pip install -r requirements.txt

4. Install Ollama and pull the phi4 model:

# Install Ollama (follow instructions for your OS)
# Pull the phi4 model
ollama pull phi4

5. Configure environment variables:

cp .env.example .env
# Edit .env with your settings

Configuration

The system uses the following environment variables:

Ollama Configuration

• OLLAMA_BASE_URL: Base URL for Ollama server (default: http://127.0.0.1:11434)
• OLLAMA_MODEL: Model name (default: phi4)
• OLLAMA_TEMPERATURE: Response creativity (default: 0.7)
• OLLAMA_TOP_P: Response diversity (default: 0.9)
• OLLAMA_NUM_CTX: Context length (default: 2048)
• OLLAMA_NUM_THREAD: Thread count (default: 4)
• OLLAMA_STOP: Stop sequences for response control
• OLLAMA_REPEAT_PENALTY: Penalty for repetitive responses
• OLLAMA_TOP_K: Top-k sampling parameter

Usage

1. Start the Ollama server:

ollama serve

2. Run the Streamlit application:

streamlit run rag_chatbot/chatbot.py

3. Upload a PDF document and start asking questions!

Project Structure

rag-chatbot/
├── docs/
│   ├── rag_flow_diagram.md
│   └── langgraph_workflow.md
├── rag_chatbot/
│   ├── chatbot.py
│   ├── models/
│   │   └── ollama_client.py
│   ├── utils/
│   │   ├── text_processor.py
│   │   ├── vector_store.py
│   │   ├── pdf_processor.py
│   │   └── logging_config.py
│   └── rag_chain.py
├── .env
├── .env.example
├── .cursorrules
├── requirements.txt
└── README.md

Contributing

1. Fork the repository
2. Create a feature branch
3. Commit your changes
4. Push to the branch
5. Create a Pull Request

License

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