About the pre-train and the adv-train

Open y52wang opened this issue 7 years ago • 1 comments

Hi Wang: I have a quesion about the Toy_dataset. In the Generator, it use self.targets to calculate pre-train loss. But, it use self.x to calculate adv-loss. Why they are different? Any hint will be appreciated. Thank you very much for your sharing.

Dec 06 '18 09:12 y52wang

Maybe the answer is hidden in your paper :-) I'm digging~

Dec 06 '18 12:12 y52wang