Fine-tuning or RAG: which one you need
What does fine-tuning change in a model, and when is it worth doing?
Fine-tuning is a little more training for a model that's already trained, on your own examples. It teaches a habit: a format, a tone, one kind of task. Try a better prompt before you reach for it. When the model needs your facts, let it look them up instead. That's RAG.
