Fetching the paper…

ConViT: Improving Vision Transformers with Soft Convolutional Inductive Biases · Around