Share your thoughts, 1 month free Claude Pro on usSee more
WorkDL logo mark

LoRA+: Efficient Low Rank Adaptation of Large Models

About

In this paper, we show that Low Rank Adaptation (LoRA) as originally introduced in Hu et al. (2021) leads to suboptimal finetuning of models with large width (embedding dimension). This is due to the fact that adapter matrices A and B in LoRA are updated with the same learning rate. Using scaling arguments for large width networks, we demonstrate that using the same learning rate for A and B does not allow efficient feature learning. We then show that this suboptimality of LoRA can be corrected simply by setting different learning rates for the LoRA adapter matrices A and B with a well-chosen ratio. We call this proposed algorithm LoRA$+$. In our extensive experiments, LoRA$+$ improves performance (1-2 $\%$ improvements) and finetuning speed (up to $\sim$ 2X SpeedUp), at the same computational cost as LoRA.

Soufiane Hayou, Nikhil Ghosh, Bin Yu• 2024

Related benchmarks

TaskDatasetResultRank
Mathematical ReasoningGSM8K
Accuracy74.42
1424
Code GenerationHumanEval
Pass@120.07
1048
Mathematical ReasoningGSM8K (test)
Accuracy52.11
816
Code GenerationHumanEval (test)
Pass@118.17
701
ClassificationCars
Accuracy82.4
571
Question AnsweringARC-E
Accuracy88.26
544
Image ClassificationVTAB 1K
Overall Mean Accuracy72.7
359
Question AnsweringOBQA
Accuracy83.2
347
Code GenerationHumanEval
pass@146.34
329
Image ClassificationPets
Accuracy94.07
320
Showing 10 of 50 rows

Other info

Follow for update