WGAN implementation error

[marued]

In your WGAN implementation, you have your n-critique loop around the Generator learning when it should actually be around the Discriminator's. Critique stands for Discriminator.

[eriklindernoren]

The paper tells us that we should train the critic (discriminator) n_critic times for each generator update, just as WGAN is implemented here. The discriminator is updated on each batch but the generator is only updated every n_critic batch.

[marued]

A yes, sorry! I read too quickly and mistook the 'If' in the code for a 'for' loop.

[eriklindernoren]

No problem! :)