Fine-tuning

In one line

Training an already-trained model a little more on your own examples so it behaves the way you need.

Picture it

A new hire who already knows the job, spending a week learning how your company does it.

Under the hood

Continued training on a smaller, task-specific dataset; full fine-tuning, LoRA adapters, instruction tuning. It changes behaviour and style reliably and adds facts unreliably. For facts, use retrieval.