AI Paper Atlas

Research map / Architectures and efficiency / Attention and transformers

Evolving Attention with Residual Convolutions

Yujing Wang, Yaming Yang, Jiangang Bai, Mingliang Zhang, Jing Bai, Jing Yu, Ce Zhang, Gao Huang, Yunhai Tong

ICML 2021 · Architectures and efficiency › Attention and transformers · approach: Attention head and block redesign · 9 citations (OpenAlex)

This paper tried a common change: Add residual connections

Add skip connections between attention maps · ImageNet · Top-1 accuracy: 74.29 -> 75.23 Improved

What 546 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
Add evolving attention to AA-ResNetAA-ResNet · ImageNetTop-1 accuracy: 74.33 -> 75.23; 77.70 -> 78.22; 78.50 -> 79.29; 79.10 -> 79.63Improved
Add evolving attentionBERT-Base · GLUEAvg: 80.9 -> 83.3Improved
Add evolving attentionT5-Base · GLUEAvg: 83.4 -> 84.5Improved
Add evolving attentionBERT-Large · GLUEAvg: 83.7 -> 85.3Improved
Add evolving attentionRoBERTa-Large · GLUEAvg: 86.4 -> 87.2Improved
Add evolving attentionTransformer-Lite and Transformer-Base · Machine translationBLUE: 33.32 -> 33.80; 21.11 -> 21.63; 33.22 -> 34.12; 27.47 -> 27.56; 40.79 -> 41.54Improved
Add attention convolution (ablation)EA-AA-ResNet-34 · ImageNetTop-1 accuracy: 74.34 -> 75.23Improved
Add skip connections between attention maps (ablation)EA-AA-ResNet-34 · ImageNetTop-1 accuracy: 74.29 -> 75.23Improved
Use 3 × 3 rather than 1 × 1 convolution (ablation)EA-AA-ResNet-34 · ImageNetTop-1 accuracy: 74.99 -> 75.23Improved
Use 3 × 3 rather than 5 × 5 convolution (ablation)EA-AA-ResNet-34 · ImageNetTop-1 accuracy: 75.12 -> 75.23Mixed
Add attention convolution and skip connections (ablation)EA-BERT-Base · CoLAscore: 52.1 -> 60.0; 52.9 -> 60.0Improved
Use 3 × 3 rather than 1 × 1 convolution (ablation)EA-BERT-Base · CoLAscore: 57.5 -> 60.0Improved

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