Figure 3 shows the generated samples with improved training of WGAN by the gradient penalty (GP-WGAN) and ours with the consistency regularization (CT-GAN), ...
確定! 回上一頁