Swin Transformer with different input size

[mt-cly]

Hi, thanks for your great work.
The Swin Transformer seems to throw an error if the input x does not satisfy the size requirement of x.W%window_size==0 and x.H%window_size==0. I wonder if it is convenient to add some code to pad some zero pixels to the input x so that images with various sizes can be directly taken as the input of Swin Transformer.
Thank you.

[rwightman]

In case anyone else reads this, for swin and other transformer arch with fixed or contrained image sizes, it's best to resize/crop/pad the image in preprocessing to fit the network input. That's already handled by default eval transforms for the validation/train scripts here.. but it's not hard to customize for padding or different types of resize (non-aspect preserving, etc).