rag-cookbooks

Why we included this project

This collection is for developers who have a basic RAG setup working and want to see what the advanced and agentic variants look like in real code. It opens with a plain naive RAG baseline, then moves through techniques like hybrid search, HyDE, query rewriting, self-querying, and agentic retrieval flows, each shown as a runnable notebook with the relevant research paper linked for follow-up reading. Since this is a learning resource rather than something you pip install, treat it as a guided tour: it explains why each technique matters and how to evaluate it before you adapt the approach to your own project. Teams working in LangChain with Pinecone, Weaviate, Qdrant, or FAISS will find the examples easy to lift and modify, and the built-in evaluation guidance helps you judge whether a technique actually improves your answers rather than just looking clever.