🤖 Chat with PDF locally using Ollama + LangChain

A powerful local RAG (Retrieval Augmented Generation) application that lets you chat with your PDF documents using Ollama and LangChain. This project includes both a Jupyter notebook for experimentation and a Streamlit web interface for easy interaction.

Project Structure

ollama_pdf_rag/
├── src/                      # Source code
│   ├── app/                  # Streamlit application
│   │   ├── components/       # UI components
│   │   │   ├── chat.py      # Chat interface
│   │   │   ├── pdf_viewer.py # PDF display
│   │   │   └── sidebar.py   # Sidebar controls
│   │   └── main.py          # Main app
│   └── core/                 # Core functionality
│       ├── document.py       # Document processing
│       ├── embeddings.py     # Vector embeddings
│       ├── llm.py           # LLM setup
│       └── rag.py           # RAG pipeline
├── data/                     # Data storage
│   ├── pdfs/                # PDF storage
│   │   └── sample/          # Sample PDFs
│   └── vectors/             # Vector DB storage
├── notebooks/               # Jupyter notebooks
│   └── experiments/         # Experimental notebooks
├── tests/                   # Unit tests
├── docs/                    # Documentation
└── run.py                   # Application runner

📺 Video Tutorial

✨ Features

• 🔒 Fully local processing - no data leaves your machine
• 📄 PDF processing with intelligent chunking
• 🧠 Multi-query retrieval for better context understanding
• 🎯 Advanced RAG implementation using LangChain
• 🖥️ Clean Streamlit interface
• 📓 Jupyter notebook for experimentation

🚀 Getting Started

Prerequisites

1. Install Ollama

   • Visit Ollama's website to download and install
   • Pull required models:

     ollama pull llama3.2  # or your preferred model
     ollama pull nomic-embed-text

2. Clone Repository

   git clone https://github.com/tonykipkemboi/ollama_pdf_rag.git
   cd ollama_pdf_rag

3. Set Up Environment

   python -m venv venv
   source venv/bin/activate  # On Windows: .\venv\Scripts\activate
   pip install -r requirements.txt

   Key dependencies and their versions:

   ollama==0.4.4
   streamlit==1.40.0
   pdfplumber==0.11.4
   langchain==0.1.20
   langchain-core==0.1.53
   langchain-ollama==0.0.2
   chromadb==0.4.22

🎮 Running the Application

Option 1: Streamlit Interface

python run.py

Then open your browser to http://localhost:8501

Streamlit interface showing PDF viewer and chat functionality

Option 2: Jupyter Notebook

jupyter notebook

Open updated_rag_notebook.ipynb to experiment with the code

💡 Usage Tips

1. Upload PDF: Use the file uploader in the Streamlit interface or try the sample PDF
2. Select Model: Choose from your locally available Ollama models
3. Ask Questions: Start chatting with your PDF through the chat interface
4. Adjust Display: Use the zoom slider to adjust PDF visibility
5. Clean Up: Use the "Delete Collection" button when switching documents

🤝 Contributing

Feel free to:

• Open issues for bugs or suggestions
• Submit pull requests
• Comment on the YouTube video for questions
• Star the repository if you find it useful!

⚠️ Troubleshooting

• Ensure Ollama is running in the background
• Check that required models are downloaded
• Verify Python environment is activated
• For Windows users, ensure WSL2 is properly configured if using Ollama

Common Errors

ONNX DLL Error

If you encounter this error:

DLL load failed while importing onnx_copy2py_export: a dynamic link Library (DLL) initialization routine failed.

Try these solutions:

1. Install Microsoft Visual C++ Redistributable:

   • Download and install both x64 and x86 versions from Microsoft's official website
   • Restart your computer after installation

2. If the error persists, try installing ONNX Runtime manually:

   pip uninstall onnxruntime onnxruntime-gpu
   pip install onnxruntime

CPU-Only Systems

If you're running on a CPU-only system:

1. Ensure you have the CPU version of ONNX Runtime:

   pip uninstall onnxruntime-gpu  # Remove GPU version if installed
   pip install onnxruntime  # Install CPU-only version

2. You may need to modify the chunk size in the code to prevent memory issues:

   • Reduce chunk_size to 500-1000 if you experience memory problems
   • Increase chunk_overlap for better context preservation

Note: The application will run slower on CPU-only systems, but it will still work effectively.

🧪 Testing

Running Tests

# Run all tests
python -m unittest discover tests

# Run tests verbosely
python -m unittest discover tests -v

Pre-commit Hooks

The project uses pre-commit hooks to ensure code quality. To set up:

pip install pre-commit
pre-commit install

This will:

• Run tests before each commit
• Run linting checks
• Ensure code quality standards are met

Continuous Integration

The project uses GitHub Actions for CI. On every push and pull request:

• Tests are run on multiple Python versions (3.9, 3.10, 3.11)
• Dependencies are installed
• Ollama models are pulled
• Test results are uploaded as artifacts

📝 License

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

---

⭐️ Star History

Built with ❤️ by Tony Kipkemboi!

Follow me on X | LinkedIn | YouTube | GitHub