Getting to know the generation confrontation network (12) -- using pytoch to build wgan-gp to generate handwritten digits

〇、WGAN-GP Introduce

WGAN-GP (Wasserstein GAN using gradient penalty), That is to say WGAN On the basis of joining gradient penalty Gradient penalty term , Please refer to the paper ：《Improved Training of Wasserstein GANs》.

One 、 generator

Generator classes and WGAN identical , The code is as follows ：

class Generator(nn.Module):
    def __init__<