lora

LoRA (Low-Rank Adaptation) is a technique used to reduce the number of parameters that need to be fine-tuned in large pre-trained models by introducing low-rank matrices for the adaptation process, making it more efficient and accessible for resource-constrained environments.

13 papers