Ways to Implement Lipschitz continuity

I’m wondering can we implement spectral normalization (SN) to the weights in the critic as well as using gradient penalty (GP) in the loss function? The authors of the SN-GAN paper were introducing it as an alternative and more robust approach, however can applying both SN and GP be of any help and is it mathematically correct?

Mohamed Hamza

Hey @Mohamed_Hamza,

I hope you are doing great. Recently, I was going through some articles regarding Lipschitz Continuity and Spectral Normalization.
Here are the links article 1, article 2. You might find these interesting!