Swapped width and height

[Hunger-Prevails]

Hi Gabriele, huge respect for what you managed to achieve here.
Your knowledge and expertise about tensor indexing truly impress as previously I couldn't imagine how random crop operations could be done in batches.
And after studying your code for a while I'm gradually coming to understand how such complex indexing works.

But one small thing here on line 123 of batchnorms.py really buffles me:

w, h = padded.size(2), padded.size(3)

I can't understand why the height and width variables are swapped. If this is indeed your intention, could you kindly shed some light on why it should be this way?

[GabPrato]

Hi, just wanted to let you know that I've seen your message. Having a quick look, it does seem like I've accidentally swapped the w and h. But I would need to deep dive back into the code to make sure. I will do that soon :)

[Hunger-Prevails]

Hi, just wanted to let you know that I've seen your message. Having a quick look, it does seem like I've accidentally swapped the w and h. But I would need to deep dive back into the code to make sure. I will do that soon :)

Hi, I tried out your implementation with w and h restored to the natural order back when I opened this issue and it functions perfectly to this day. I think you were probably working with square-shaped inputs when you wrote this, therefore this accidental swap didn't pose a problem for you.

Anyway, your implementation saves me a lot of time and allows my code to run quite a bit faster!

[GabPrato]

You were right about the issue. Thank you very much for pointing that out. It indeed did not matter if H and W were swapped for squared inputs and crashed when input was not square. It's fixed now! Happy Holidays 😄