AI Paper Atlas

Research map / Architectures and efficiency / Attention and transformers

Vision-TTT: Efficient and Expressive Visual Representation Learning with Test-Time Training

Quan Kong, Yanru Xiao, Yuhao Shen, Cong Wang

ECCV 2026 · Architectures and efficiency › Attention and transformers · approach: Hybrid and interpretable ViT designs · 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
share Q / K projections in TTT layer (ablation)Vittt-T · ImageNet-1KTop-1 accuracy: 74.2 -> 74.0No clear effect
add gating (ablation)Vittt-T · ImageNet-1KTop-1 accuracy: 74.0 -> 75.2Improved
add Conv1d (vanilla TTT) (ablation)Vittt-T · ImageNet-1KTop-1 accuracy: 75.2 -> 75.6Improved
add gating and Conv1d together (ablation)Vittt-T · ImageNet, COCO, ADE20KAcc / AP^b / mIoU / throughput: +1.6 acc, +1.5 AP^b, +1.4 mIoU, -14.1% throughputMixed
add dual dataset strategy (forward and backward) (ablation)Vittt-T · ImageNet-1KTop-1 accuracy: 75.6 -> 77.5Improved
add dual dataset strategy (ablation)Vittt-T · COCO2017 / ADE20KAP^b / mIoU: 42.7 vs 41.6 AP^b; 44.1 vs 42.6 mIoUImproved
add dual dataset strategy (ablation)Vittt-T · ThroughputFPS: 3346 -> 2068Hurt
add Conv2d dataset preprocessing (ablation)Vittt-T · ImageNet-1KTop-1 accuracy: 77.5 -> 77.7Improved
use MeanPool instead of head class token (ablation)Vittt-T · ImageNet-1KTop-1 accuracy: 76.6 -> 77.7Improved
use MeanPool instead of DoubleClassTok (ablation)Vittt-T · ImageNet-1KTop-1 accuracy: 75.4 -> 77.7Improved
use MeanPool instead of MaxPool (ablation)Vittt-T · ImageNet-1KTop-1 accuracy: 77.2 -> 77.7Improved
vary mini-batch size b from 4 to 196 (ablation)Vittt-T · ImageNet-1KTop-1 accuracyMixed

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