AI Paper Atlas

Research map / Architectures and efficiency / Attention and transformers

Faster Vision Transformers with Adaptive Patches

Rohan Choudhury, JungEun Kim, Jinhyung Park, Eunho Yang, Laszlo A. Jeni, Kris Kitani

ICLR 2026 · Architectures and efficiency › Attention and transformers · approach: Hybrid and interpretable ViT designs

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
use zero-initialized connection instead of a residual connection (ablation)APT · ImageNet-1k (with training)accuracy: 87.52 -> 88.13Improved
use zero-initialized connection instead of a non-zero initialized connection (ablation)APT · ImageNet-1k (with training)accuracy: 87.81 -> 88.13Improved
use zero-initialized connection instead of a residual connection (ablation)APT · ImageNet-1k (without training)accuracy: 87.40 -> 87.98Improved
use adaptive patches instead of random token dropping (ablation)ViT-B MAE 384/16 · ImageNet-1k full fine-tuningaccuracy: 83.4 -> 84.2Improved
use adaptive patches instead of resizing only (ablation)ViT-L MAE 336/14 · ImageNet-1k full fine-tuningaccuracy: 85.9 -> 86.1Improved
use adaptive patches instead of resizing only (ablation)EVA-02-L · ADE20K 512mIoU: 58.81 -> 59.70Improved
use adaptive patches instead of resizing only (ablation)LLaVA-1.5-13B · VQAv2accuracy: 78.9 -> 79.4Improved
apply APT to ViT-B at 384/16 for full fine-tuningViT-B MAE · ImageNet-1ktraining time: 11.6h -> 9.0h; +29%Improved
apply APT to ViT-L at 448/14ViT-L MAE · ImageNet-1ktraining time: 31.4h -> 16.9h; +86%Improved
apply APT with no token reduction (tau = -1) (ablation)ViT-B 224/16 · ImageNet-1k throughputImg/s: 3310 -> 3090Hurt
apply APT with no token reduction (tau = -1) (ablation)ViT-L 336/14 · ImageNet-1k throughputImg/s: 395 -> 360Hurt
increase entropy threshold (ablation)APT · ImageNet-1kaccuracyHurt

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