This is a Retrieval Augmented Generation (RAG) application built with LangChain, OpenAI, and Streamlit. It allows users to upload PDF documents and ask questions about their content.

• 📄 PDF document upload and processing
• 🔍 Semantic search using ChromaDB vector store
• 🤖 GPT-3.5-turbo powered responses
• 🎨 Clean and intuitive Streamlit UI
• ⚙️ Configurable model parameters (temperature, number of retrieved chunks)
• 📊 Document chunking with overlap for better context

• Python 3.9 or higher
• OpenAI API key

1. Clone the repository:

git clone <repository-url>
cd rag-tutorial

2. Create a virtual environment:

python3 -m venv venv
source venv/bin/activate  # On Windows: venv\Scripts\activate

3. Install dependencies:

pip install -r requirements.txt

4. Create a .env file in the project root:

OPENAI_API_KEY=your_openai_api_key_here

5. Run the Streamlit app:

streamlit run app.py

The app will open in your browser at http://localhost:8501

1. Push your code to GitHub
2. Go to https://share.streamlit.io/
3. Click "New app"
4. Select your GitHub repository, branch, and app.py file
5. Click "Deploy"
6. Add your OPENAI_API_KEY in the Streamlit Cloud secrets manager (Settings → Secrets)

1. Create a new Space on Hugging Face

2. Select "Docker" as the SDK

3. Push the following files to your Space repository:

   • app.py - Main Streamlit application
   • requirements.txt - Python dependencies
   • Dockerfile - Docker configuration
   • .env - Environment variables (add via Space settings/secrets)

4. The Space will automatically build and deploy

For security, add your OpenAI API key as a secret:

1. Go to your Space settings
2. Click "Repository secrets"
3. Add OPENAI_API_KEY with your OpenAI API key value

rag-tutorial/
├── app.py                          # Main Streamlit application
├── rag-notebook.ipynb              # Jupyter notebook with RAG implementation
├── requirements.txt                # Python dependencies
├── Dockerfile                      # Docker configuration
├── README.md                       # This file
├── .env                           # Environment variables (not in git)
├── .env.example                   # Example environment variables
└── chroma_db/                     # Vector store persistence directory

1. Retrieval: The app searches the ChromaDB vector store to find the most relevant document chunks based on the user's query.

2. Augmentation: The retrieved chunks are added to the prompt as context.

3. Generation: The GPT-3.5-turbo model uses this context to generate an accurate, grounded answer.

• Temperature (0.0 - 1.0): Controls creativity vs. determinism

  • Lower values: More focused, deterministic responses
  • Higher values: More creative, varied responses

• Number of Retrieved Chunks (k) (1 - 10): How many document chunks to retrieve for context

  • Lower values: Faster but potentially less context
  • Higher values: More context but may dilute focus

• streamlit - Web app framework
• langchain - LLM framework
• langchain-openai - OpenAI integration
• langchain-community - Community integrations (PDF loader, Chroma)
• chromadb - Vector database
• openai - OpenAI API client
• pypdf - PDF processing
• python-dotenv - Environment variable management

• Ensure you've installed requirements: pip install -r requirements.txt
• Check your virtual environment is activated

• Create a .env file in the project root with: OPENAI_API_KEY=your_key
• On Hugging Face Spaces, add it as a secret in space settings

• Ensure the PDF is not corrupted
• Check file size (very large PDFs may cause issues)
• Verify OpenAI API key is valid

• Reduce the number of retrieved chunks (k parameter)
• This speeds up processing at the cost of less context

• Support for multiple file formats (DOCX, TXT, etc.)
• Batch document processing
• Chat history and session management
• Different LLM model options
• Advanced filtering and keyword search
• Response confidence scoring

MIT License

For issues or questions, please create an issue in the GitHub repository.