Browse Teach Curate

Low-Rank Adaptation (LoRA)

Parameter-efficient fine-tuning by learning low-rank weight deltas on frozen pretrained weights; why rank-limited updates suffice and how adapters merge at inference.

Demerzel Video

Low-Rank Adaptation (LoRA)

Parameter-efficient fine-tuning by learning low-rank weight deltas on frozen pretrained weights; why rank-limited updates suffice and how adapters merge at inference.