← Tutorials20 / 22
What is an LLM?
What is RAG (retrieval-augmented generation)?Tap to reveal
question ─▶ embed ─▶ vector search
│ top-k chunks
▼
prompt = question + chunks ─▶ LLM
│
grounded answerInstead of relying on what the model memorized, you fetch relevant documents at question time and put them in the prompt.
Typically: split documents into chunks, embed each chunk as a vector, store them in a vector index, retrieve the chunks nearest to the question, then ask the model to answer from them.