Is this code applicable to large batch size?

[Guptajakala]

You mentioned you are training with batch size 1 in this case. I need to run on a dataset where the images are much smaller and my GPU allows the batch size to be up to 200. Will the adversarial training work in this case?

[hfslyc]

We train with batch size 1 only because of memory limitation. The method ideally works with any batch size. But with different dataset, the hyperparameters or network architecture could be tuned for better performance.