Glossary · Model Training

Full Fine-Tuning

Updating all of a model's parameters during training, as opposed to parameter-efficient methods like LoRA that only train a small added subset — more resource-intensive but sometimes necessary for large behavior shifts.