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.


Ship secure code faster

Crash Override integrates security into the developer workflow. No context switching, no waiting on reviews.