Glossary

Fine-Tuning

Short answer

Training an existing model on new data to specialise it for a particular domain.

Fine-tuning is the far cheaper alternative to training a model from scratch: you train on your own data on top of an existing model.

In corporate use the typical goal is brand consistency: a model fine-tuned on a brand's entire visual archive generates in that brand's visual language.

LoRA is fine-tuning's light and cheap form; full fine-tuning gives stronger results but demands far more data, compute and expertise.