Fine-tuning
Further training of an existing model on your own data to shape its behaviour.
Fine-tuning takes a trained model and runs more training on a narrower dataset: your own writing style, your label taxonomy, your domain. The result is a model that behaves closer to how you want without a long prompt explaining it every time.
It is not a knowledge transfer mechanism. Fine-tuning teaches behaviour and format, not facts you expect the model to look up. If you want the model to know your documentation, use retrieval. Fine-tuning also costs money: a training run burns compute, and quality depends on having enough clean examples.
Reach for it when prompting has failed repeatedly and the behaviour is stable. Do not reach for it because a demo felt generic.
Related terms
Training
The expensive process that sets a model's parameters by exposing it to data.
Dataset
A collection of examples used to train, evaluate, or ground a system.
RAG
Fetch the relevant documents first, then let the model answer using them.
Prompt Engineering
The craft of writing instructions a model follows reliably.
Parameters
The learned numbers inside a model. Their count is the model size.
Want this working inside your team?
It starts with one conversation about your workflow.