Closed tianjiu233 closed 4 years ago
In your codes, the fcn traincoah there is a line that outputs = net(inputs*masks).detach()
If the detach() is added, the backward() will not work. And the only loss will only be KLD. Is is right or there is something else I do not know?
Hope for your reply!
In your codes, the fcn traincoah there is a line that outputs = net(inputs*masks).detach()
If the detach() is added, the backward() will not work. And the only loss will only be KLD. Is is right or there is something else I do not know?
Hope for your reply!