What does retrieval-augmented generation (RAG) add to an LLM call?
- Retrieved documents in the prompt
- Extra layers trained on the fly
- A larger tokenizer vocabulary
- A second model that votes on output
RAG searches a knowledge source and inserts relevant passages into the context to ground the answer.
Play AI & ML ▸