AI Paper Atlas

Research map / Architectures and efficiency / Attention and transformers

FLatten Transformer: Vision Transformer using Focused Linear Attention

Dongchen Han, Xuran Pan, Yizeng Han, Shiji Song, Gao Huang

ICCV 2023 · Architectures and efficiency › Attention and transformers · approach: Efficient and robust vision transformers · 394 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 focused function to vanilla linear attention (ablation)FLatten-DeiT-T · ImageNet-1K valTop-1 accuracy (%): 70.5 -> 71.8Improved
add depthwise convolution (DWC) rank restoration (ablation)FLatten-DeiT-T · ImageNet-1K valTop-1 accuracy (%): 71.8 -> 74.1Improved
vary focused factor p between 2 and 32 (ablation)FLatten-Swin-T · ImageNet-1K valTop-1 accuracy (%): 82.03, 82.11, 81.94, 81.99, 81.88No clear effect
enlarge attention window from 7^2 to 56^2 (ablation)FLatten-Swin-T · ImageNet-1K valTop-1 accuracy (%): 81.6 -> 82.1Improved
apply focused linear attention to first two stages instead of only stage 1 (ablation)FLatten-Swin-T · ImageNet-1K valTop-1 accuracy (%): 81.7 -> 82.1Improved
also apply focused linear attention in last two stages (ablation)FLatten-Swin-T · ImageNet-1K valTop-1 accuracy (%): 82.1 -> 81.9Hurt
use FLatten linear attention instead of other linear attention designsDeiT-T setting · ImageNet-1K valTop-1 accuracy (%): 72.9 -> 74.1Improved
replace Softmax attention with FLatten in DeiT-TDeiT-T · ImageNet-1K valTop-1 accuracy (%): 72.2 -> 74.1Improved
replace Swin attention with FLattenSwin-T · ADE20K UperNetmIoU: 44.51 -> 44.82Improved
replace attention with FLatten in larger models (CSwin-B at 384)CSwin-B · ImageNet-1K valTop-1 accuracy (%): 85.4 -> 85.5No clear effect

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