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

LoCO: Low-rank Compositional Rotation Fine-tuning

About

Parameter-efficient fine-tuning (PEFT) has emerged as an critical technique for adapting large-scale foundation models across natural language processing and computer vision. While existing methods such as low-rank adaptations achieve parameter efficiency via low-rank weight updates, they are limited in their ability to preserve the geometric structure of pretrained representations. We introduce Low-rank Compositional Orthogonal fine-tuning (LoCO), a novel PEFT method that constructs orthogonal transformations through low-rank skew-symmetric matrices and compositional rotation chains. We propose an approximation scheme that enables fully parallel computation of compositional rotations, making the approach practical for high-dimensional feature spaces. Our method maintains low computational complexity while maintaining orthogonality with controlled approximation error. We validate LoCO across diverse domains, including diffusion transformer fine-tuning, vision transformer adaptation, and language model adaptation. Our method demonstrates superior or competitive performance compared to both existing orthogonal and non-orthogonal methods.

An Nguyen, Jaesik Choi, Anh Tong• 2026

Related benchmarks

TaskDatasetResultRank
Image ClassificationVTAB 1k (test)--
121
Mathematical ReasoningGSM8K (val)
Accuracy50.19
108
Mathematical ReasoningMATH (val)
Accuracy8.4
59
Natural Language UnderstandingGLUE (val)
CoLA Score65.56
26
Controllable Image GenerationCanny Controllable Generation Benchmark
F1 Score49
3
Controllable Image GenerationDepth Controllable Generation Benchmark
Mean Squared Error (MSE)723
3
Controllable Image GenerationMask Controllable Generation Benchmark
MSE6.93e+3
3
Controllable Image GenerationColorization Controllable Generation Benchmark
MSE106
3
Controllable Image GenerationDeblur Controllable Generation Benchmark
MSE83
3
Showing 9 of 9 rows

Other info

Follow for update