[5] An Image is Worth 16x16 Words: Transformers for Image Recognition at ScaleViT backbone 2021Q1 re-read