Low-Rank Adaptation

An adaptation method for LLMs.

Pros

• unlike Adapters, no additional Inference latency.

References

• LoRA: Low-Rank Adaptation of Large Language Models