AI glossary
What is fine-tuning in AI?
Fine-tuning means further training an already trained model on a small set of your own examples, to specialize it in a task, a style or a format.
Training an LLM from scratch is extremely expensive. Fine-tuning starts from a model that already knows the language and adjusts its weights with hundreds or thousands of examples of what you want it to do.
It is useful to lock in a tone, an output format or a very repetitive task. To give it new knowledge that changes often, RAG is usually the better fit, since it updates without retraining.
Before fine-tuning, it is worth checking whether a good prompt with examples solves the problem: it often does, and it is much cheaper.
An example
A company that wants the model to sort support tickets into its own categories can fine-tune it on tickets that are already sorted.