Fine-tuning

Fine-tuning further trains an existing model on additional examples to adapt its behavior, tone, or format. It changes the model's weights, unlike retrieval, which supplies information at query time. Fine-tuning suits consistent behavior; retrieval suits current facts.

The two are frequently confused in procurement conversations. If the requirement is that answers reflect this week's policy, fine-tuning is the wrong instrument, because the model would need retraining whenever the policy changes.

If the requirement is that outputs consistently follow a house format or specialized vocabulary, fine-tuning does what prompting struggles to sustain across long or varied workloads.