AI Paper Atlas

Research map / Architectures and efficiency / Attention and transformers

Vision Transformer Finetuning Benefits from Non-Smooth Components

Ambroise Odonnat, Laetitia Chapel, Romain Tavenard, Ievgen Redko

ICML 2026 · Architectures and efficiency › Attention and transformers · approach: Vision transformer design · 0 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
finetune attention (MHA) instead of LN1 (ablation)ViT-Base SGD · 11 benchmarks avgtop-1 accuracy (%): 89.8 -> 90.8Improved
finetune FC1 instead of LN1 (ablation)ViT-Base SGD · 11 benchmarks avgtop-1 accuracy (%): 89.8 -> 90.7Improved
finetune FC2 instead of LN1 (ablation)ViT-Base SGD · 11 benchmarks avgtop-1 accuracy (%): 89.8 -> 90.3Improved
finetune MHA instead of LN2 (ablation)ViT-Base SGD · 11 benchmarks avgtop-1 accuracy (%): 89.9 -> 90.8Improved
finetune MHA instead of FC1 (ablation)ViT-Base SGD · 11 benchmarks avgtop-1 accuracy (%): 90.7 -> 90.8No clear effect
finetune MHA instead of LN1 (ablation)ViT-Base Adam · 4 datasets avgtop-1 accuracy (%): 79.2 -> 82.2Improved
finetune MHA instead of LN2 (ablation)ViT-Base Adam · 4 datasets avgtop-1 accuracy (%): 79.4 -> 82.2Improved
finetune MHA instead of FC1 (ablation)ViT-Base Adam · 4 datasets avgtop-1 accuracy (%): 81.4 -> 82.2Improved
finetune MHA instead of LN1 (ablation)ViT-Base low-data · 4 datasets avgtop-1 accuracy (%): 93.0 -> 94.4Improved
finetune MHA instead of FC1 (ablation)ViT-Base low-data · 4 datasets avgtop-1 accuracy (%): 93.8 -> 94.4Improved
finetune LayerNorms only instead of LoRA (ablation)ViT-Base Adam · Cifar100top-1 accuracy (%): 88.1 -> 89.4Mixed
finetune FC1 instead of MHA (ablation)ViT-Base Adam · Cifar100top-1 accuracy (%): 91.0 -> 91.3No clear effect

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy