RAG-Assistant

A RAG chatbot which can processes PDF files

🤖 RAG Chatbot

A Retrieval-Augmented Generation (RAG) chatbot built with Streamlit, LangChain, and Ollama. Upload PDF documents and ask questions about their content!

✨ Features

• PDF Document Processing: Upload and process multiple PDF files
• Smart Text Chunking: Intelligent document splitting for better retrieval
• Vector Search: Fast semantic search using ChromaDB and sentence transformers
• RAG Integration: Combines document retrieval with AI responses
• Modern UI: Clean, intuitive Streamlit interface
• Chat History: Persistent conversation memory

🚀 Quick Start

Prerequisites

1. Python 3.8+ installed on your system
2. Poetry package manager installed
3. Ollama installed and running locally

Installation

1. Clone or download this repository

2. Download the Gemma3 model:

   ollama pull gemma3:1b

3. Upgrade Poetry:

   pip install poetry --upgrade

4. Install dependencies:

   poetry install --no-root

5. Activate the virtual environment:

   poetry shell

6. Run the application:

   streamlit run app.py

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

📖 How to Use

1. Upload Documents

• Use the sidebar to upload one or more PDF files
• Click "Process Documents" to create embeddings
• Wait for the processing to complete

2. Ask Questions

• Type your questions in the chat interface
• bot will search through your documents and provide relevant answers
• You can ask general questions too (when no documents are uploaded)

3. Chat Features

• Clear Chat: Reset conversation history
• Document Status: See if documents are ready for querying
• Real-time Processing: Live feedback during document processing

🔧 Technical Details

Architecture

• Frontend: Streamlit web interface
• LLM: Ollama with Gemma3:1b model
• Vector Database: ChromaDB with inmemeory storage
• Embeddings: Sentence Transformers (all-MiniLM-L6-v2)
• Text Processing: Recursive character text splitting

File Structure

RAG-Assistant
├── app.py          # Main Streamlit application 
├── utils
      ├── datastore.py  # Document processing and vector DB creation
      ├── retriver.py   # Document retrieval and search
      ├── memory_cleanup.py  # Delete created embeddings
├── pyproject.toml   # Python dependencies
└── README.md         # This file

Key Components

datastore.py

• Handles PDF file uploads and processing
• Creates text chunks with optimal overlap
• Generates and stores embeddings in ChromaDB

retriver.py

• Performs semantic search on stored documents
• Returns relevant document chunks with metadata
• Includes similarity scoring and filtering

app.py

• Main Streamlit interface
• Manages chat flow and RAG integration
• Handles user interactions and responses

🐛 Troubleshooting

Common Issues

1. Ollama Connection Error

   • Ensure Ollama is running: ollama serve
   • Check if gemma3:1b model is downloaded: ollama list

2. PDF Processing Errors

   • Verify PDF files are not corrupted
   • Check file permissions and size limits

3. Memory Issues

   • Large PDFs may require more RAM
   • Consider processing smaller documents first

4. Dependency Issues

   • Update pip: pip install --upgrade pip
   • Install dependencies individually if needed

Performance Tips

• Document Size: Optimal chunk size is 1000 characters with 200 overlap
• Model Selection: Gemma3:1b provides good balance of speed and quality
• Batch Processing: Process multiple documents together for efficiency

🔒 Security Notes

• Documents are processed locally on your machine
• No data is sent to external services (except Ollama if using remote)
• Temporary files are automatically cleaned up

📝 License

This project is open source and available under the MIT License.

🤝 Contributing

Feel free to submit issues, feature requests, or pull requests to improve this RAG chatbot!

---

Happy Document Chatting! 📚💬