LoRA Adapter
A trainable low-rank decomposition matrix injected into transformer attention layers during model tuning.
LoRA (Low-Rank Adaptation) inserts rank-decomposition matrices alongside frozen weight layers. During training, only these lightweight adapters are updated, reducing memory footprints and simplifying multi-task deployment.
Historical figures and technical concepts for informational purposes only. Not technical, professional, legal, or financial advice. Sources: Official Documentation.