This work targets to merge various Vision Transformers (ViTs) trained on different tasks (ie,
datasets with different object categories) or domains (ie, datasets with the same categories …

Recent works have demonstrated that transformer can achieve promising performance in
computer vision, by exploiting the relationship among image patches with self-attention …

Vision transformers (ViTs) have been successfully applied in image classification tasks
recently. In this paper, we show that, unlike convolution neural networks (CNNs) that can be …

Vision transformers (ViTs) achieve remarkable performance on large datasets, but tend to
perform worse than convolutional neural networks (CNNs) when trained from scratch on …

Vision transformers are known to be more computationally and data-intensive than CNN
models. These transformer models such as ViT, require all the input image tokens to learn …

Conclusion In this study, we proposed a novel method called UP-ViTs to prune ViTs in a
unified manner. Our framework can prune all components in a ViT and its variants, maintain …

Patch-based models, eg, Vision Transformers (ViTs) and Mixers, have shown impressive
results on various visual recognition tasks, alternating classic convolutional networks. While …

Vision transformers (ViTs) have gained increasing popularity as they are commonly believed
to own higher modeling capacity and representation flexibility, than traditional convolutional …

Transformers increasingly dominate the machine learning landscape across many tasks and
domains, which increases the importance for understanding their outputs. While their …

Vision Transformers (ViTs) take all the image patches as tokens and construct multi-head
self-attention (MHSA) among them. Complete leverage of these image tokens brings …