AI Paper Atlas

Research map / Architectures and efficiency / Attention and transformers

Patch Slimming for Efficient Vision Transformers

Yehui Tang, Kai Han, Yunhe Wang, Chang Xu, Jianyuan Guo, Chao Xu, Dacheng Tao

CVPR 2022 · Architectures and efficiency › Attention and transformers · approach: Efficient and robust vision transformers · 185 citations (OpenAlex)

Check 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
Apply static top-down patch slimmingDeiT-Ti · ImageNetTop-1 accuracy: 72.2 -> 72.0Hurt
Apply dynamic top-down patch slimmingDeiT-Ti · ImageNetFLOPs reduction: 53.8Improved
Apply static top-down patch slimmingDeiT-S · ImageNetTop-1 accuracy: 79.8 -> 79.4Hurt
Apply dynamic top-down patch slimmingDeiT-B · ImageNetTop-1 accuracy: 81.8 -> 81.6Hurt
Apply patch slimmingT2T-ViT-14 · ImageNetFLOPs: 5.2 -> 3.1Improved
Use learned, layer-specific patch selection instead of uniform pruning (ablation)DeiT-S · ImageNetTop-1 accuracy: 77.2 -> 79.4Improved
Increase the global tolerant error to allow more patches to be pruned (ablation)DeiT-S · ImageNetFLOPs reductionImproved
Prune patches in deeper layers rather than the same fraction in shallower layers (ablation)DeiT-S · ImageNetTop-1 accuracyImproved
Select patches using propagated impact scores instead of random or current-layer attention scores (ablation)DeiT-S · ImageNetTop-1 accuracyImproved

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

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