Fine-Tuning
Fine-tuning continues training a pretrained model on your own examples so it gets better at a specific task, tone, or format. It changes the model's weights, unlike prompting or RAG, which change what you feed it.
Fine-tuning takes a model that already knows language and trains it further on a focused set of your own examples, adjusting its weights so it’s better at your particular task, voice, or output format. It’s different in kind from prompting (changing the instruction) and from RAG (changing the context you supply): those leave the model as-is, while fine-tuning changes the model itself.
It’s worth it when you need consistent behavior on a narrow, repeated task and you have enough quality examples to teach it — and especially when a smaller fine-tuned model can replace a larger general one at a fraction of the cost. It’s the wrong tool for keeping a model up to date on changing facts; that’s what retrieval is for. The honest trade-off is effort and freshness: fine-tuning is slower to set up and goes stale when the task shifts, so reach for it when the win is specialization, not knowledge.
