tags from_paper computer_vision attention machine_learning How Do Vision Transformers Work Data Augmentation: