How to train your ViT? Data, Augmentation, and Regularization in Vision Transformers - 42Papers