AI Paper Atlas

Research map / Architectures and efficiency / Attention and transformers

DropPos: Pre-Training Vision Transformers by Reconstructing Dropped Positions

Haochen Wang, Junsong Fan, Yuxi Wang, Kaiyou Song, Tong Wang, Zhaoxiang Zhang

NeurIPS 2023 · Architectures and efficiency › Attention and transformers · approach: Efficient and robust vision transformers · 4 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
use patch mask ratio 0.75 instead of 0 (whole image visible) (ablation)DropPos · ImageNet-1K fine-tunetop-1: 81.94 -> 82.96Improved
use patch mask ratio 0.75 instead of 0 (ablation)DropPos · ADE20kmIoU: 31.66 -> 40.68Improved
raise mask ratio from 0.75 to 0.90 (ablation)DropPos · ADE20kmIoU: 40.68 -> 40.30Hurt
set position mask ratio to 1.00 instead of 0.75 (no reference patch) (ablation)DropPos · ImageNet-1K fine-tunetop-1: 82.96 -> 82.66Hurt
set position mask ratio to 1.00 instead of 0.75 (ablation)DropPos · ADE20kmIoU: 40.68 -> 39.98Hurt
remove position smoothing (sigma=0) versus linear decay 1->0 (ablation)DropPos · COCO detectionAP^b: 40.32 -> 42.14Improved
use constant sigma=2 instead of decay 1->0 (ablation)DropPos · ImageNet-1K fine-tunetop-1: 82.79 -> 82.96Hurt
remove attentive reconstruction (tau=inf) (ablation)DropPos · COCO detectionAP^b: 40.56 -> 42.14Improved
remove attentive reconstruction (tau=inf) (ablation)DropPos · ADE20kmIoU: 38.38 -> 40.68Improved
use tau=0.5 instead of 0.1 (ablation)DropPos · ADE20kmIoU: 40.68 -> 38.80Hurt
make sin-cos position embedding learnable (ablation)DropPos · ADE20kmIoU: 40.68 -> 39.37Hurt
random learnable position embedding instead of fixed sin-cos (ablation)DropPos · ImageNet-1K fine-tunetop-1: 82.96 -> 82.48Hurt

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