Some questions about code detail

[JianqiangRen]

When calculate loss_vae_gan, why try to minimize the difference between D_real and 0.9? why not 1.0?

[youngwoon]

It is one of the GAN training techniques called label smoothing which reduces the networks' vulnerability (see Section 3.4. in https://arxiv.org/abs/1606.03498 for more details).