Vision transformers (ViTs) have achieved remarkable performance in various computer vision tasks. However, they demand significant computational and memory bandwidth, making them difficult to ...