Open daifeng2016 opened 4 years ago
Hi, based on the theory of GAN, the optimization of Discriminator is done first, followed by Generator optimization. However, Generator is always trained first in your code. Does it make sense?
其实没啥关系,换个角度,你同时看两个batch,中间那部分是不是相当于就是先判别器,后生成器了,其实都不影响的.
Hi, based on the theory of GAN, the optimization of Discriminator is done first, followed by Generator optimization. However, Generator is always trained first in your code. Does it make sense?