Abstract: The advent of Vision Transformers (ViT) has set a new performance leap in computer vision by leveraging self-attention mechanisms. However, the computational efficiency of ViTs is limited by ...