🎥 YouTube Video Q&A with RAG

Ask questions about YouTube videos using Retrieval-Augmented Generation (RAG) powered by Hugging Face models.

✨ Features

• Smart Q&A: Ask questions about any YouTube video content
• Multiple URL Formats: Supports full URLs or just video IDs
• AI-Powered: Uses Mistral-7B and sentence transformers for accurate responses
• User-Friendly: Clean Streamlit interface with real-time feedback
• Flexible Configuration: Environment-based settings

🚀 Quick Start

1. Manual Setup

# Create environment
conda create -n rag_pipeline python=3.10 -y
conda activate rag_pipeline

# Install dependencies
conda install numpy scipy scikit-learn -y
pip install -r requirements.txt

2. Configure Environment

Create a .env file:

HUGGINGFACE_TOKEN=your_token_here

Get your token from Hugging Face Settings

3. Run Application

streamlit run rag_pipeline.py

🎯 How It Works

1. Input: Paste a YouTube URL or video ID
2. Processing: Extracts transcript and creates vector embeddings
3. Query: Ask your question about the video content
4. Answer: Get AI-powered responses based on the transcript

📋 Supported URL Formats

• https://www.youtube.com/watch?v=VIDEO_ID
• https://youtu.be/VIDEO_ID
• https://www.youtube.com/embed/VIDEO_ID
• Just the video ID: VIDEO_ID

⚙️ Configuration

Customize settings in .env:

• HUGGINGFACE_TOKEN: Your HF authentication token
• CHUNK_SIZE: Text processing chunk size (default: 800)
• RETRIEVAL_K: Number of relevant chunks to retrieve (default: 4)
• MAX_NEW_TOKENS: Maximum response length (default: 512)

🛠️ Technologies Used

• Frontend: Streamlit
• LLM: Mistral-7B-Instruct via Hugging Face
• Embeddings: sentence-transformers/all-MiniLM-L6-v2
• Vector Store: FAISS
• Framework: LangChain

🔧 Troubleshooting

Common Issues

• No transcript available: Some videos don't have captions
• Token errors: Ensure your HF token has proper permissions
• Installation issues: Use the automated setup script

System Requirements

• Python 3.10+
• 4GB+ RAM recommended
• Internet connection for model downloads

📝 Example Questions

• "What are the main topics discussed in this video?"
• "Can you summarize the key points?"
• "What examples were given about [specific topic]?"
• "Who is mentioned in the video?"

Note: This tool only works with videos that have available transcripts/captions.