GAN网络走过的坑

所用语言：PyTorch，python3.6
复现DCGAN代码时出现ERROR如下：
Trying to backward through the graph a second time, but the buffers have already been freed. Specify retain_graph=True when calling backward the first time.

后在所有的loss中添加retain_graph=True，解决了该问题。
loss_d.backward(retain_graph=True)
但伴随着出现了如下问题
out of memory
在gen_data后加detach()，删除上面添加的retain_graph=True，解决了上述所有问题

d_fake = discriminator(gen_data.detach())

参考文献

[1] Pytorch: detach 和 retain_graph