Open druzkaya opened 5 years ago
In vanilla GANs, a sigmoid activation is applied on the output layer for G and D. See https://github.com/goodfeli/adversarial/blob/master/mnist.yaml
For WGAN, there is none for D, and we get a score instead of a probability.
However, in the code there is no activation (e.g sigmoid) for WGAN-MLP also for G, whereas there is tanh for WGAN-DCGAN. Is there a specific reason?
Thanks in advance
In vanilla GANs, a sigmoid activation is applied on the output layer for G and D. See https://github.com/goodfeli/adversarial/blob/master/mnist.yaml
For WGAN, there is none for D, and we get a score instead of a probability.
However, in the code there is no activation (e.g sigmoid) for WGAN-MLP also for G, whereas there is tanh for WGAN-DCGAN. Is there a specific reason?
Thanks in advance