zju-vipa / osgan Goto Github PK

Does this make gradient direction of G and D the same ?

I changed the adversarial loss to hinge gan loss, and gamma will often be -inf in the training loop, then the training will be crash.
How can I adjust the "get_gradient_ratios" to prevent gamma from being -inf?

Thanks for sharing your code and it's really a great work.
Can your technique be applied to PatchGAN?

Thanks for sharing your code and it's really a great work.

I would like to apply this technique to other models using GANs to reduce the training time.
For implementation, I have two questions about your code.

1. Why do you set reduction='none' and get the average from loss_fake in one stage ?
   train_dcgan_asymmetric_one_stage_simple.py#L60
2. How do you extract the gradient of the generator from the gradient of the Discriminator?
   In this code train_dcgan_asymmetric_one_stage_simple.py#L72, I think that the generator backpropagates the same gradient as the discriminator.

Thank you.