Models, reasoning & training · Concept

LoRA

Also called: low-rank adaptation

Low-rank adaptation: learn low-rank parameter updates instead of updating all model weights.