About training process.

[Jamie-Cheung]

In stylediffusion, the content image and style image are both used in training. But why is only the style image used in your main.py?

[mac-mvak]

Hello, I'm very pleased, that you as the author of the paper created an issue in my repo. I have a hardcoded path to the imagenet dataset, from which I sample pictures. I would like to change it further and add a parameter for a custom path.

I would appreciate it if you as an author of the paper checked the correctness of the losses, which I use in my code, because I might have missed some tricky nuances.