Open dougsouza opened 5 years ago
@dougsouza do you implement it??
@Johnson-yue, yes. I tried gradient penalty only in the real/fake output of the network. But I didn't pursue much, I went back to spectral normalization.
Maybe spectral normalization is in common use
Gradient penalty claculation should change when D has an auxiliary classifier or the gradient penalty should only be calculated over the real/fake output? Is there any intuition behind of whatever is the best solution?
Thanks!