Skip to content
ai

LoRA

Low-Rank Adaptation

Definition

LoRA is a parameter-efficient fine-tuning method that injects trainable low-rank decomposition matrices into a frozen pre-trained model's weight matrices. Instead of updating all weights, only the small rank-decomposition matrices (typically 0.1–1% of parameters) are trained.

LoRA enables high-quality fine-tuning on a single GPU with minimal memory overhead and has become the dominant PEFT technique.

Can this ship? Are we exposed? Can you prove it?

Every CISO is asking these three questions. We answer them with facts, not guesses, from repository to runtime.