AI Paper Atlas

Research map / Architectures and efficiency / Attention and transformers

Learning Expressive Prompting With Residuals for Vision Transformers

Rajshekhar Das, Yonatan Dukler, Avinash Ravichandran, Ashwin Swaminathan

CVPR 2023 · Architectures and efficiency › Attention and transformers · approach: Vision transformer design · 20 citations (OpenAlex)

This paper tried a common change: Add attention module

Add residual prompts at attention LayerNorm · FGVC · accuracy: +3.5% Improved

What 492 accepted papers report for this change

See this change in detailMatched by the wording of each paper's own result, read by AI from the full text. Papers report gains more often than failures, so this shows what gets published, not the chance a change works.

Read the paper

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 EXPRES instead of VPT-deepViT-B/16 adaptation · VTAB-1k natural meanaccuracy: 78.48 -> 79.7Improved
Use EXPRES instead of FT-allViT-B/16 adaptation · VTAB-1k specialized meanaccuracy: 83.36 -> 84.0Improved
Use EXPRES instead of VPT-deepViT-B/16 adaptation · VTAB-1k structured meanaccuracy: 54.98 -> 55.0No clear effect
Use EXPRES instead of VPT-deepViT-B/16 few-shot segmentation · PASCAL-5i 5-shotmIoU: 33.4 -> 41.8; 56.2 -> 60.2; 49.8 -> 52.4; 47.7 -> 51.4Improved
Use EXPRES instead of VPT-deepViT-B/16 adaptation · Oxford Flowersaccuracy: 99.0 -> 99.0No clear effect
Use EXPRES instead of VPT-deepViT-B/16 adaptation · CUBaccuracy: 88.5 -> 88.3Hurt
Use EXPRES instead of VPT-deepViT-B/16 adaptation · Stanford Carsaccuracy: 83.6 -> 80.5Hurt
Use last-layer keys rather than queries or MLP outputs (ablation)EXPRES for few-shot segmentation · PASCAL-5i fold 0 train-split, 5-shotaccuracy: 44.39 -> 51.61; 38.25 -> 51.61Improved
Allow prompt interactions with other tokens through more encoder layers (ablation)EXPRES · FGVC and VTAB prompt propagation experimentsaccuracyImproved
Add residual prompts at attention LayerNorm (ablation)EXPRES · FGVCaccuracy: +3.5%Improved
Add residual prompts to query-key-value projections (ablation)EXPRES · FGVCaccuracy: +2.8%Improved
Prompt the attention block rather than the MLP block (ablation)EXPRES · FGVCaccuracy: +1.6%Improved

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