A fine-tuning method for generative AI models, like Adam or LoRA, that fine-tune only parts of a model.
"PEFT let us adjust the model for legal terminology without changing the underlying model, for a lot less compute cost and review time."
