pytorch-beginner icon indicating copy to clipboard operation
pytorch-beginner copied to clipboard

issues about GAN

Open 978749951 opened this issue 6 years ago • 1 comments

@L1aoXingyu 你好,GAN网络的训练判别器的时候不是要把生成器固定住吗?但是,代码中并没有哪一步把生成器的参数固定住。这样在训练判别器的时候,生成器的参数也会进行梯度更新的吧?

978749951 avatar Jul 25 '18 12:07 978749951

因为我的optimizer里面只放入了判别器的参数,所以不固定生成器,生成器也不会更新,当然更好的是把生成器的输出 detach 出来

On Jul 25, 2018, at 20:49, Jinchang Chen [email protected] wrote:

GAN网络的训练判别器的时候不是要把生成器固定住吗?

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/L1aoXingyu/pytorch-beginner/issues/12, or mute the thread https://github.com/notifications/unsubscribe-auth/ASWpwJJHW7damhuSWJ84xEbW1ypvN5vXks5uKGlEgaJpZM4VgBYJ.

L1aoXingyu avatar Aug 01 '18 03:08 L1aoXingyu