The gan loss is negative？

[Hans1984]

Hello, I clone your code and try the demo training, however, I found the gan loss of generator and discriminator without GP is negative. Is that normal? And the total aeu-loss doesn't seem to converge. Did I make a mistake somewhere？ I appreciate for any help.

[Yang7879]

Hi @Hans1984 , according to the definition, the loss is negative. It indeed takes some time to converge at the beginning, you are suggested to track it over a few thousand iterations.

[Hans1984]

Thanks a lot, however, the training process uses mass memory resources, lol.