Neural Mastery

Training

Fine-tuning

Fine-tuning continues training a pretrained model on a narrower objective or dataset.

Pretrained modelAdd your examplesTrain at a low learning rateEvaluate on held-out data
Fine-tuning continues training a pretrained model on a narrower objective or dataset. Values shown in the worked visual are illustrative unless the labels state otherwise.

How it works

Pretraining supplies broad representations, and fine-tuning adapts behavior to a domain or task. Updating all weights costs more memory than parameter-efficient methods.

A concrete example

Illustrative: adapt a 7-billion-parameter model on 50,000 instruction-response pairs.

Do not confuse it with Prompting

fine-tuning changes parameters; prompting changes only the request context.

Study the full lesson