RAG_Techniques

Why we included this project

RAG_Techniques is a community-maintained collection of runnable notebooks that starts with the basics of retrieval-augmented generation and works up to agentic and hybrid designs. Each notebook explains the reasoning behind a technique, then shows working code and references, so you can follow along and adapt the pattern to your own pipeline. Teams outgrowing a simple stuff-the-context prototype will find the material on query rewriting, reranking, hybrid search, and evaluation especially useful. Because this is a curated index rather than a single installable library, treat it as a learning path and a source of ideas, then lift the techniques you need into your own stack. With dozens of contributors and an active community, it stays current as RAG keeps evolving.