Losses explanation

[ghost]

Could someone provide e briefly explanation of the different losses implemented?
In particular:
'G_GAN', 'G_GAN_Feat', 'G_VGG', 'D_real', 'D_fake'

Thanks

[tcwang0509]

Losses starting with 'G' are generator losses, while losses starting with 'D' are discriminator losses.
'G_GAN': generator GAN loss
'G_GAN_Feat': feature matching loss in discriminator layers
'G_VGG': feature matching loss in VGG layers
'D_real': discriminator loss on real images
'D_fake': discriminator loss on fake images