A chatbot designed to help understand scientific papers: feed it a document, and it answers questions grounded in that document instead of making things up.
The idea (RAG, Retrieval-Augmented Generation): the document is chunked and indexed into a vector store, then the most relevant passages are retrieved and injected into the language model’s context before it answers — which sharply reduces hallucinations and keeps answers anchored to the source text.
Stack
- LangChain — orchestrates the RAG pipeline
- Chroma — vector store for retrieving relevant passages
- Ollama (Llama 3.2) — language model, run locally
- Streamlit — chat interface