AI Paper Atlas

Research map / Architectures and efficiency / Attention and transformers

Sliced Recursive Transformer

Zhiqiang Shen, Zechun Liu, Eric Xing

ECCV 2022 · Architectures and efficiency › Attention and transformers · approach: Vision transformer pruning · 24 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 non-linear projection layer (NLL) between recursions (ablation)recursive transformer · ImageNet-1KTop-1: 72.5 -> 74.7Improved
add learnable residual connection (LRC) (ablation)Recursion + NLL · ImageNet-1KTop-1: 74.7 -> 75.2Improved
add spatial pyramid stem (ablation)Recursion + NLL · ImageNet-1KTop-1: 74.7 -> 76.0Improved
remove class token (ablation)Recursion + NLL · ImageNet-1KTop-1: 74.7 -> 75.0Improved
use all components together (ablation)Recursion + NLL · ImageNet-1KTop-1: 74.7 -> 76.8Improved
soft distillation instead of GT + hard distillation (ablation)SReT-TL · ImageNet-1KTop-1: 77.5 -> 77.9Improved
use group self-attention with recursion instead of global attention without loop (ablation)SReT-ExT · ImageNet-1KTop-1: 73.6 -> 74.0Improved
use internal loops instead of external loop (ablation)12-layer recursive transformer · ImageNet-1KTop-1: 67.0 -> 67.6Improved
use group self-attention approximation [8,8][4,4][1,1] (ablation)SReT-TL · ImageNet-1KTop-1: 76.78 -> 76.06Hurt
use fewer groups [8,2][4,1][1,1] (ablation)SReT-TL · ImageNet-1KTop-1: 76.06 -> 76.65Improved
add inverse permutation layer after sliced group attention (ablation)SReT-TL group design · ImageNet-1KTop-1: 75.41 -> 75.94Improved
add distillationSReT-T · ImageNet-1KTop-1: 76.0 -> 77.6Improved

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