Zhenliang He

@yingcong Thanks for the code! Have you tried Facelet Bank for other attributes such as bangs and eyeglasses?

@LMescheder Hi, most GAN papers use Adam optimizer. What's the main reason here using RMSProp? Is it better than Adam here?

@takerum The paper derives the projection discriminator from binary cross entropy loss with sigmoid activation. However, hinge loss with linear activation is used in the experiment, is there any theoretical...