Mar 11, 2021
Yes. The original paper called it "constant" because it used to be variable latent factors in traditional GAN. The paper claims it is not necessary and replaces it with learned parameters that do not change during inference.
Yes. The original paper called it "constant" because it used to be variable latent factors in traditional GAN. The paper claims it is not necessary and replaces it with learned parameters that do not change during inference.
