We attempt to reduce the computational costs in vision transformers (ViTs), which increase quadratically in the token number. We present a novel training paradigm that trains only one ViT model at a time, but is capable of providing improved image recognition performance with various computational costs. Here, the trained ViT model, termed super vision transformer (SuperViT), is empowered with the versatile ability to solve incoming patches of multiple sizes as well as preserve informative tokens with multiple keeping rates (the ratio of keeping tokens) to achieve good hardware efficiency for inference, given that the available hardware resources often change from time to time. Experimental results on ImageNet demonstrate that our SuperViT can considerably reduce the computational costs of ViT models with even performance increase. For example, we reduce 2x FLOPs of DeiT-S while increasing the Top-1 accuracy by 0.2% and 0.7% for 1.5x reduction. Also, our SuperViT significantly outperforms existing studies on efficient vision transformers. For example, when consuming the same amount of FLOPs, our SuperViT surpasses the recent state-of-the-art (SOTA) EViT by 1.1% when using DeiT-S as their backbones. The project of this work is made publicly available at https://github.com/lmbxmu/SuperViT.

翻译：我们试图降低视觉Transformer（ViTs）在token数量上呈二次增长的计算成本。我们提出了一种新颖的训练范式，该范式每次仅训练一个ViT模型，但能够以不同的计算成本提供改进的图像识别性能。在此，所训练的ViT模型被称为超级视觉Transformer（SuperViT），它被赋予多方面的能力：能够处理多种尺寸的输入图像块，并以多种保留率（保留token的比例）保留信息丰富的token，从而在推理时根据时常变化的可用硬件资源实现良好的硬件效率。在ImageNet上的实验结果表明，我们的SuperViT能够显著降低ViT模型的计算成本，甚至提升性能。例如，我们将DeiT-S的FLOPs降低2倍，同时使Top-1准确率提升0.2%，当降低1.5倍时准确率提升0.7%。此外，我们的SuperViT显著优于现有的高效视觉Transformer研究。例如，当消耗相同的FLOPs时，以DeiT-S为骨干网络，我们的SuperViT比近期最先进（SOTA）的EViT高出1.1%。本工作的项目已公开发布于https://github.com/lmbxmu/SuperViT。