I’m trying to implement aspect ratio bucketing to be able to train with images of seemingly all resolutions, but I can’t manage to find the training sample size (height and width) to edit to make this work.
I started from the text_to_image example in the diffusers documentation.
Thanks in advance