AI Paper Atlas

Research map / Architectures and efficiency / Attention and transformers

RNN as Linear Transformer: A Closer Investigation into Representational Potentials of Visual Mamba Models

Timing Yang, Feng Wang, Guoyizhe Wei

CVPR 2026 · Architectures and efficiency › Attention and transformers · approach: Sparse and structured attention · 1 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
replace linear attention with Mamba-2 mixerDINO ViT-B · ImageNet-1k linear probingtop-1 accuracy: 74.7 -> 78.1Improved
replace softmax self-attention with Mamba-2 mixerDINO ViT-B · ImageNet-1k linear probingtop-1 accuracy: 78.2 -> 78.1No clear effect
replace linear attention with Mamba-2 mixerDINO ViT-B · ADE20KmIoU: 29.2 -> 38.0Improved
replace softmax self-attention with Mamba-2 mixerDINO ViT-B · ADE20KmIoU: 43.2 -> 38.0Hurt
replace linear attention with Mamba-2 mixerDINO ViT-B · COCO 2017box AP: 37.1 -> 42.8Improved
replace softmax self-attention with Mamba-2 mixerDINO ViT-B · COCO 2017box AP: 44.8 -> 42.8Hurt
replace linear attention with Mamba-2 mixerDINO ViT-B · ImageNet-Atop-1 accuracy: 9.6 -> 14.2Improved
replace softmax self-attention with Mamba-2 mixerDINO ViT-B · ImageNet-Sketchtop-1 accuracy: 25.5 -> 27.6Improved
add register tokens (Mamba-Reg) and heavier headDinoVim-B · ImageNet-1k linear probingtop-1 accuracy: 78.3 -> 78.5Improved
use self-supervised DINO instead of supervised trainingVision backbones · feature-map Binary-AUCAUCImproved
increase model sizeDinoVim · ImageNet-1k linear probingtop-1 accuracy: 73.7 -> 77.4Improved

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