AI Paper Atlas

Research map / Architectures and efficiency / Attention and transformers

TAB: Transformer Attention Bottlenecks enable User Intervention and Debugging in Vision-Language Models

Pooyan Rahmanzadehgervi, Hung Huy Nguyen, Rosanne Liu, Long Mai, Anh Totti Nguyen

ICCV 2025 · Architectures and efficiency › Attention and transformers · approach: Theory of self-attention dynamics · 3 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
add attention supervision (ablation)TAB4IDC (ViT-B/16) · OpenImages-I captioningCIDEr: 324.4 -> 335.6Improved
add attention supervision (ablation)TAB (ViT-B/16) · OpenImages-I localization (PG+)PG+ mean accuracy: 61.73 -> 99.19Improved
add attention bottleneck (TAB) in place of CLIP4IDC MHSA (ablation)CLIP4IDC (ViT-B/16) · OpenImages-I localization (PG+)PG+ mean accuracy: 52.98 -> 99.19Improved
replace MHSA with TABCLIP4IDC (ViT-B/16) · CLEVR-Change localization (PG+)PG+ mean accuracy: 84.47 -> 96.75Improved
replace MHSA with TAB (ViT-B/32)CLIP4IDC · CLEVR-Change captioningCIDEr: 150.7 -> 149.6No clear effect
replace MHSA with TAB (ViT-B/16)CLIP4IDC · CLEVR-Change captioningCIDEr: 144.7 -> 153.0Improved
replace MHSA with TAB (ViT-B/32)CLIP4IDC · STD captioningBERTScore: 29.4 -> 22.6Hurt
replace MHSA with TAB (ViT-B/16)CLIP4IDC · STD captioningBERTScore: 23.0 -> 28.3Improved
use higher-resolution ViT-B/16 instead of B/32 (ablation)TAB · OpenImages-I localization (PG+)PG+ mean accuracy: 98.11 -> 99.19Improved
replace attention map with groundtruth (CorrectAttention) (ablation)TAB4IDC · OpenImages-I editing (attention supervised, ViT-B/16)object name accuracy: 88.92 -> 91.49Improved
replace attention map with groundtruth (CorrectAttention) (ablation)TAB4IDC (CLEVR-Change) · CLEVR-Change editingchange accuracy: 94.30 -> 100.0Improved
replace attention map with zeros (ZeroAttention) (ablation)TAB4IDC · OpenImages-I editing (attention supervised, ViT-B/16)change accuracy: 99.93 -> 0.0Hurt

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