convolution-vision-transformers

convolution-vision-transformers copied to clipboard

Hi,I have a question as followed:In the paper,the stride of depthwise convolution in convolution projection is 2,but in your repo stride is 1,I want to know stride is 2 or...

qdd1234

Whats the acc of this implementation on ImageNet ?

1

Thanks for releasing your CVT implementation. Can you recurrent the accuracy in the paper?

go2sea

When will you release code and scripts for training CVT on imagenet from scratch?

1

We are looking forward to it.

zhaoguangxiang