AI
Learn retrieval-augmented generation
Retrieval-augmented generation fetches relevant documents at question time and gives them to a model so the answer can be grounded in those sources.
For engineers who want to build a question-answering system over their own documents, not only chat with a model.
Start here
Read the Pinecone overview, then build the Haystack tutorial on a handful of documents you can read yourself.
Recommended resources
Start here · BeginnerWhat is retrieval-augmented generation?
A direct explanation of retrieve-then-generate and where embeddings, the index, and the model each sit.
www.pinecone.ioBuild · IntermediateYour first RAG pipeline
A complete tutorial that indexes documents and answers questions from them.
haystack.deepset.aiBuild · IntermediateRAG cookbook
A second implementation, useful for seeing which pieces stay the same when the library changes.
huggingface.coPaper · AdvancedRetrieval-Augmented Generation for Knowledge-Intensive NLP Tasks
The paper that defined RAG. Read it after you have seen a pipeline, so the method has something to attach to.
arxiv.orgRelated topics
Evaluate a RAG system · Choose and use a vector database · Understand transformer models · Learn LangGraph