A complete RAG system for biomedical literature that compares three answering approaches side-by-side.
- Downloads papers from PubMed Central
- Semantic search with sentence-transformers
- 3-way output comparison: Raw Retrieval | Standard LLM | Few-shot LLM
- Open in Colab
- Add your Groq API key
- Run all cells
- Raw Retrieval: Shows similarity scores (0.534, 0.291, 0.289)
- Standard LLM: Direct mutation list
- Few-shot LLM: Structured answer with percentages (EGFR 61.5%, KRAS 8.2%, etc.)