Should i scale input image

[Jserax]

Hello, @kjsman, thanks for this easily readable implementation. I have a question: am I correct that I should scale the input with a sampler before passing the image to the U-Net model during the training process?