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

Parameter-Efficient Fine-Tuning with Learnable Rank

About

Low-Rank Adaptation (LoRA) is a popular parameter-efficient fine-tuning (PEFT) method that restricts weight updates to low-rank adapters, introducing a fixed low-rank inductive bias by optimizing in a low-dimensional subspace. In this work, we question whether a fixed-rank constraint is the most effective inductive bias for parameter-efficient fine-tuning. We introduce *Learnable Rank LoRA (LR-LoRA)*, a PEFT method in which the adapter rank is learned during the training process. Instead of prescribing a uniform rank for all adapter layers, LR-LoRA allows the optimizer to determine the appropriate rank for each layer. Using this approach, we find substantial layer-wise variation in the learned ranks, with the attention and MLP layers in the transformer models exhibiting systematically different rank preferences. Across a range of language understanding and commonsense reasoning benchmarks, LR-LoRA achieves state-of-the-art performance in most settings and consistently outperforms strong PEFT baselines, demonstrating that a learnable rank provides a more flexible and effective inductive bias than fixed-rank adaptations.

Arpit Garg, Simon Lucey, Hemanth Saratchandran• 2026

Related benchmarks

TaskDatasetResultRank
Image ClassificationCIFAR-100--
362
ClassificationImageNet 1k (test val)
Top-1 Accuracy79.6
142
Image ClassificationSUN397
Top-1 Accuracy70.2
129
Commonsense ReasoningCommonsense 8 Sub-Tasks
Accuracy (8 Sub-Tasks)88.22
80
Natural Language UnderstandingGLUE
SST-2 Accuracy96.1
62
Instruction TuningMT-Bench
Score6.55
16
Image ClassificationCIFAR-100 (test val)
Top-1 Accuracy85.4
4
Image ClassificationFlowers102 (test val)
Top-1 Accuracy96.1
4
Image ClassificationSUN397 (test val)
Top-1 Accuracy68.1
4
Showing 9 of 9 rows

Other info

Follow for update