Abstract: Vision Transformers have achieved high performance in computer vision tasks, but their high computational cost and low throughput are weaknesses. Therefore, much research has been done to ...