Serving
Retrieval-augmented generation
Also called RAG.
What is retrieval-augmented generation?
Retrieval-augmented generation fetches relevant documents at request time and places them in the model’s context. It supplies facts that change; fine-tuning supplies behaviour that should not. The two are complements, not alternatives.
The choice is often framed as either-or, which is a category error. Retrieval decides what the model sees; fine-tuning decides what it does with what it sees. A system usually wants both.
Fine-tuning improves retrieval systems specifically by teaching the model to use context faithfully and to decline when the retrieved material does not support an answer — which is the behaviour that makes such a system trustworthy.
Related terms
Start with the free tier
A magic link creates your account, your tenant and your first API key. No card until you ask for compute.