AI Paper Atlas

Research map / Architectures and efficiency / Attention and transformers

Thicker and Quicker: The Jumbo Token for Fast Plain Vision Transformers

Anthony Fuller, Yousef Yassin, Daniel Kyrollos, Evan Shelhamer, James R Green

ICLR 2026 · Architectures and efficiency › Attention and transformers · approach: Hybrid and interpretable ViT designs

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 Jumbo token instead of registersViT-Nano · ImageNet-1K val (224px)top-1 (%): 61.0 -> 74.6Improved
add Jumbo token instead of registers, larger widthViT-Base · ImageNet-1K val (224px)top-1 (%): 84.3 -> 84.9Improved
add Jumbo token instead of registersViT-Small · ImageNet-21Ktop-1 (%): 41.5 -> 44.61Improved
add Jumbo to MAE pretrainingViT-Base · ImageNet-1K linear probetop-1 (%): 68.1 -> 73.0Improved
add Jumbo for test-time adaptation with SARViT-Small · ImageNet-Cavg accuracy (%): 54.9 -> 60.1Improved
add Jumbo to fine-tuning for segmentationViT-Base · ADE20KmIoU (%): 42.5 -> 44.4Improved
do not share Jumbo FFN across layers (ablation)Jumbo ViT-Small · ImageNet-21Ktop-1 (%): 44.61 -> 44.95Improved
add layer-specific LoRA to shared Jumbo FFN (ablation)Jumbo ViT-Small · ImageNet-21Ktop-1 (%): 44.61 -> 44.94Improved
add Jumbo FFN (ablation)Jumbo ViT-Small · ImageNet-21Ktop-1 (%): 43.64 -> 44.61Improved
increase Jumbo width multiplier J from 6 to 10 (ablation)Jumbo ViT-Small · ImageNet-21Ktop-1 (%): 44.61 -> 45.62Improved
widen registers ViT-Base to match Jumbo parameters (ablation)ViT-Base+Registers · ImageNet-21Ktop-1 (%): 45.7 -> 47.1No clear effect
add separate FFN for registers (ablation)ViT-Small+Registers · ImageNet-21Ktop-1 (%): 41.5 -> 42.1No 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