Definition: low-rank adaptation

A method for fine-tuning a large language model (LLM). Low-rank adaptation (LoRA) trains a small add-on rather than retraining the entire model. See AI fine tuning.