A comprehensive chatbot that teaches Formula 1 racing from both sporting and engineering perspectives using Retrieval-Augmented Generation (RAG) with free/open-source models.
- Comprehensive Knowledge Base: Covers F1 rules, regulations, history, engineering, and technical aspects
- RAG Implementation: Uses semantic search to retrieve relevant context for accurate responses
- Free Models: Built with HuggingFace transformers and free models (no API costs)
- Interactive Interface: User-friendly Streamlit web interface
- Educational Focus: Designed specifically for beginners learning about F1
- Install dependencies:
pip install -r requirements.txt- Run the chatbot:
streamlit run app.py- Knowledge Base: Structured F1 data in markdown files
- Vector Store: ChromaDB for semantic search
- Embeddings: Sentence transformers for document and query embeddings
- LLM: HuggingFace transformers (free models like GPT-2, T5, or Llama-2)
- Interface: Streamlit for web-based chat interface
The chatbot can answer questions about:
- F1 rules and regulations
- Race procedures and formats
- Engineering and technical aspects
- Historical information
- Driver and team information
- Car specifications and aerodynamics
- Knowledge Base Accuracy: Vector search and retrieval needs optimization for better relevance
- Response Intelligence: Chatbot responses need enhancement for better educational value
- LLM Integration: Disabled due to PyTorch security vulnerability (CVE-2025-32434)
- Content Quality: Some retrieved content may not be optimally relevant to queries
- Vector similarity scoring optimization
- Response generation improvement (currently using basic fallback)
- Knowledge base chunking strategy refinement
- F1 content filtering and relevance scoring
- Integration with secure LLM models
- Some RSS feeds may be intermittently unavailable
- API endpoints (Ergast) occasionally down
- Content deduplication needed
- Source attribution consistency
🟡 Beta Version - Functional but requires debugging and optimization for production use