Applied AI·Fine-tuning and customisation

prompting plateaued, so you take a few hundred of your own examples and train the model on them until it does the thing by default.

Fine-tuning

Draft summary, pending review

Continuing training on your own examples so behaviour moves toward them. It teaches style, format and task-specific behaviour well; it is a poor way to inject facts, which retrieval handles better and more updatably.

Commonly confused with

Retrieval-augmented generation (RAG)

Fine-tuning vs Retrieval-augmented generation (RAG) →