Closed fsfwl11406 closed 2 years ago
loss_encoder = self.adv_cost(dis_logits, 1 - dis_labels)这行代码的这个loss有什么用
Hi,
这个就是用来训encoder的loss,和它相针对的是discriminator的loss,discriminator只在训练过程中有用(对抗),在测试阶段就丢掉了。
loss_encoder = self.adv_cost(dis_logits, 1 - dis_labels)这行代码的这个loss有什么用