Simpler Fast Vision Transformers with a Jumbo CLS Token

We introduce a simple enhancement to the global processing of vision transformers (ViTs) to improve accuracy while maintaining throughput. Our approach, Jumbo, creates a wider CLS token, which is split to match the patch token width before attention, processed with self-attention, and reassembled. After attention, Jumbo applies a dedicated, wider FFN to this token. Jumbo significantly improves over ViT+Registers on ImageNet-1K at high speeds (by 3.2% for ViT-tiny and 13.5% for ViT-nano); these Jumbo models even outperform specialized compute-efficient models while preserving the architectural advantages of plain ViTs. Although Jumbo sees no gains for ViT-small on ImageNet-1K, it gains 3.4% on ImageNet-21K over ViT+Registers. Both findings indicate that Jumbo is most helpful when the ViT is otherwise too narrow for the task. Finally, we show that Jumbo can be easily adapted to excel on data beyond images, e.g., time series.

LookHere: Vision Transformers with Directed Attention Generalize and Extrapolate

High-resolution images offer more information about scenes that can improve model accuracy. However, the dominant model architecture in computer vision, the vision transformer (ViT), cannot effectively leverage larger images without finetuning -- ViTs poorly extrapolate to more patches at test time, although transformers offer sequence length flexibility. We attribute this shortcoming to the current patch position encoding methods, which create a distribution shift when extrapolating. We propose a drop-in replacement for the position encoding of plain ViTs that restricts attention heads to fixed fields of view, pointed in different directions, using 2D attention masks. Our novel method, called LookHere, provides translation-equivariance, ensures attention head diversity, and limits the distribution shift that attention heads face when extrapolating. We demonstrate that LookHere improves performance on classification (avg. 1.6%), against adversarial attack (avg. 5.4%), and decreases calibration error (avg. 1.5%) -- on ImageNet without extrapolation. With extrapolation, LookHere outperforms the current SoTA position encoding method, 2D-RoPE, by 21.7% on ImageNet when trained at $224^2$ px and tested at $1024^2$ px. Additionally, we release a high-resolution test set to improve the evaluation of high-resolution image classifiers, called ImageNet-HR.