31. CvT : Introducing Convolutions to Vision Transformer