MODEL ADAPTATION • 227 / 397
Know when to prompt, retrieve, fine-tune or compress a model.
LoRA
LoRA is a parameter-efficient fine-tuning technique that learns small low-rank adapter matrices instead of changing every base-model weight.
Think of it like
Think of LoRA as changing how a trained worker behaves or how efficiently that worker can be deployed, rather than giving it a live reference book.
Real life
Organizations adapt general models using techniques such as LoRA.
SRE lens
It reduces training memory and storage compared with full fine-tuning.
Remember thisLoRA is a parameter-efficient fine-tuning technique that learns small low-rank adapter matrices instead of changing every base-model weight.