[D] Transfer learning on GANs?
Sorry if this question has been asked a million times before but I failed to find a good explanation so far.
Transfer learning is common for image classification task with models pre-trained from Imagenet. But how to do that for image generation? Given the recent amazing results on GAN research to generate high quality images, such as BigGAN, StyleGAN, etc., it would be ideal if I can leverage these pre-trained weights for my own small dataset.