Canonical Rank Adaptation: An Efficient Fine-Tuning Strategy for Vision Transformers

Lokesh Veeramacheneni, Moritz Wolter, Hilde Kuehne, Juergen Gall
Proceedings of the 42nd International Conference on Machine Learning, PMLR 267:61108-61125, 2025.

Abstract

Modern methods for fine-tuning a Vision Transformer (ViT) like Low-Rank Adaptation (LoRA) and its variants demonstrate impressive performance. However, these methods ignore the high-dimensional nature of Multi-Head Attention (MHA) weight tensors. To address this limitation, we propose Canonical Rank Adaptation (CaRA). CaRA leverages tensor mathematics, first by tensorising the transformer into two different tensors; one for projection layers in MHA and the other for feed-forward layers. Second, the tensorised formulation is fine-tuned using the low-rank adaptation in Canonical-Polyadic Decomposition (CPD) form. Employing CaRA efficiently minimizes the number of trainable parameters. Experimentally, CaRA outperforms existing Parameter-Efficient Fine-Tuning (PEFT) methods in visual classification benchmarks such as Visual Task Adaptation Benchmark (VTAB)-1k and Fine-Grained Visual Categorization (FGVC).

Cite this Paper

BibTeX
@InProceedings{pmlr-v267-veeramacheneni25a, title = {Canonical Rank Adaptation: An Efficient Fine-Tuning Strategy for Vision Transformers}, author = {Veeramacheneni, Lokesh and Wolter, Moritz and Kuehne, Hilde and Gall, Juergen}, booktitle = {Proceedings of the 42nd International Conference on Machine Learning}, pages = {61108--61125}, year = {2025}, editor = {Singh, Aarti and Fazel, Maryam and Hsu, Daniel and Lacoste-Julien, Simon and Berkenkamp, Felix and Maharaj, Tegan and Wagstaff, Kiri and Zhu, Jerry}, volume = {267}, series = {Proceedings of Machine Learning Research}, month = {13--19 Jul}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v267/main/assets/veeramacheneni25a/veeramacheneni25a.pdf}, url = {https://proceedings.mlr.press/v267/veeramacheneni25a.html}, abstract = {Modern methods for fine-tuning a Vision Transformer (ViT) like Low-Rank Adaptation (LoRA) and its variants demonstrate impressive performance. However, these methods ignore the high-dimensional nature of Multi-Head Attention (MHA) weight tensors. To address this limitation, we propose Canonical Rank Adaptation (CaRA). CaRA leverages tensor mathematics, first by tensorising the transformer into two different tensors; one for projection layers in MHA and the other for feed-forward layers. Second, the tensorised formulation is fine-tuned using the low-rank adaptation in Canonical-Polyadic Decomposition (CPD) form. Employing CaRA efficiently minimizes the number of trainable parameters. Experimentally, CaRA outperforms existing Parameter-Efficient Fine-Tuning (PEFT) methods in visual classification benchmarks such as Visual Task Adaptation Benchmark (VTAB)-1k and Fine-Grained Visual Categorization (FGVC).} }
Endnote
%0 Conference Paper %T Canonical Rank Adaptation: An Efficient Fine-Tuning Strategy for Vision Transformers %A Lokesh Veeramacheneni %A Moritz Wolter %A Hilde Kuehne %A Juergen Gall %B Proceedings of the 42nd International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2025 %E Aarti Singh %E Maryam Fazel %E Daniel Hsu %E Simon Lacoste-Julien %E Felix Berkenkamp %E Tegan Maharaj %E Kiri Wagstaff %E Jerry Zhu %F pmlr-v267-veeramacheneni25a %I PMLR %P 61108--61125 %U https://proceedings.mlr.press/v267/veeramacheneni25a.html %V 267 %X Modern methods for fine-tuning a Vision Transformer (ViT) like Low-Rank Adaptation (LoRA) and its variants demonstrate impressive performance. However, these methods ignore the high-dimensional nature of Multi-Head Attention (MHA) weight tensors. To address this limitation, we propose Canonical Rank Adaptation (CaRA). CaRA leverages tensor mathematics, first by tensorising the transformer into two different tensors; one for projection layers in MHA and the other for feed-forward layers. Second, the tensorised formulation is fine-tuned using the low-rank adaptation in Canonical-Polyadic Decomposition (CPD) form. Employing CaRA efficiently minimizes the number of trainable parameters. Experimentally, CaRA outperforms existing Parameter-Efficient Fine-Tuning (PEFT) methods in visual classification benchmarks such as Visual Task Adaptation Benchmark (VTAB)-1k and Fine-Grained Visual Categorization (FGVC).
APA
Veeramacheneni, L., Wolter, M., Kuehne, H. & Gall, J.. (2025). Canonical Rank Adaptation: An Efficient Fine-Tuning Strategy for Vision Transformers. Proceedings of the 42nd International Conference on Machine Learning, in Proceedings of Machine Learning Research 267:61108-61125 Available from https://proceedings.mlr.press/v267/veeramacheneni25a.html.

Related Material