我目前正在使用Pytorch训练GAN,以产生用于我的研究的组织病理学数据。我对生成器和鉴别器都使用BCE标准。该网络能够生成高质量的图像,但是损耗曲线对我来说有点神秘。 发生器和鉴别器损耗曲线看起来像精确的镜像。请参阅随附的张量板片段。有人可以告诉我为什么会这样吗?
编辑1:发生器损耗和鉴别器损耗曲线都应显示收敛性,对吧?
答案 0 :(得分:0)
The training curve you produce is somehow standard when training a GAN. The Generator and Discriminator are going to converge. If you plot the Gen Loss and Dis Loss together, you'll find out adversarial property. In fact, most of the time, validating the model by looking at the generated image is an efficient way. Here是我的一些作品供您参考。