Lars Mescheder

We haven't tried to combine spetral norm with our method, but depending on the spectral norm implementation your problems might be related to how our code handles trainable parameters. [Here](https://github.com/LMescheder/GAN_stability/blob/1c8fe0d22fc023f7690bca19929cc8d6aa58875e/gan_training/train.py#L38)...

Yes, `lsun_categories_train` and `lsun_categories_test` are ignored when not training on LSUN. `nlabels=1` simply indicates that there are no categories for celebA. For your last question: we currently only have a...

The moving average has a significant influence with batch normalization, as it simply ignores the learned weights from the BN when copying the model. Can you try to deactivate the...

The weight average reduces the noise but it also works without it. That said, for more varied data sets like Imagenet, the noise can be quite big. You can alternatively...