README.md

metadata

title: Rag Chatbot
emoji: 🚀
colorFrom: red
colorTo: red
sdk: docker
app_port: 8501
tags:
  - streamlit
pinned: false
short_description: 'Retrieval-Augmented Generation (RAG) chatbot '
license: mit

PDF-Based RAG Chatbot

A simple, 100% free Retrieval-Augmented Generation (RAG) chatbot that answers questions from PDF documents. No API keys required!

🔗 Links

• Live Demo: [Deploy to get your link]
• GitHub: [Your repository link]

After deployment, update these links with your actual URLs!

✨ Features

• ✅ Upload any two PDF documents
• ✅ Ask questions about the content
• ✅ 100% Free - No API keys needed
• ✅ Privacy-friendly - Everything runs locally
• ✅ Uses open-source Hugging Face models
• ✅ Fast vector search with FAISS

🚀 How to Use

Online (Hugging Face Spaces)

1. Visit the deployed app
2. Upload two PDF files
3. Click "Process PDFs" (takes ~30 seconds first time)
4. Ask questions about the documents!

Local Setup

1. Clone this repository:

git clone <your-repo-url>
cd <repo-name>

2. Install dependencies:

python setup.py

Or if you prefer:

pip install -r requirements.txt

3. Run the app:

streamlit run app.py

4. Open your browser to http://localhost:8501

Note: If you encounter dependency errors, see INSTALLATION.md for troubleshooting.

🛠️ How It Works

1. PDF Reading: Extract text from PDFs using PyPDF2
2. Text Chunking: Split documents into 1000-character chunks with 200 overlap
3. Embeddings: Convert chunks to vectors using Sentence Transformers
4. Vector Search: Store in FAISS index for fast similarity search
5. Question Answering:
   • Your question is converted to a vector
   • Top 3 most similar chunks are retrieved
   • FLAN-T5 generates an answer from the context

💻 Tech Stack

• Streamlit: Simple, clean web interface
• PyPDF2: PDF text extraction
• Sentence Transformers: Text embeddings (all-MiniLM-L6-v2)
• FAISS: Fast vector similarity search
• FLAN-T5: Answer generation (google/flan-t5-base)

All models are free and open-source from Hugging Face!

📦 Deployment to Hugging Face Spaces

1. Create a new Space on huggingface.co/spaces
2. Choose "Streamlit" as the SDK
3. Upload these files:
   • app.py
   • requirements.txt
   • README.md
4. The Space will automatically build and deploy!

💡 Example Questions

• What are the main topics in these documents?
• Summarize the key findings
• What does the document say about [specific topic]?
• List the important points mentioned

🎯 Why This Stack?

• Streamlit: Much simpler than Gradio, easy to understand
• PyPDF2: Straightforward PDF reading
• No API Keys: Everything runs locally, completely free
• Fast: FAISS provides instant search results
• Open Source: All models from Hugging Face

License

MIT