Ideas about describing each GAN

[DonaldTsang]

As noted in kozistr/Awesome-GANs#14 and kozistr/Awesome-GANs#18 I would like to request help in regards to collecting GANs. If it is possible I would like to ask for help as to:

• formalize the meta-structure of GANs outside of specific generators and discriminators (should have)
• visualize the meta-structures of GANs in a way that is easy to understand (see pic below) (should have)
• turn the GAN meta-structure into a Python class that can allow for hot-swappable generators and discriminators (could have)
• update Awesome-GANs with more GAN models and architectures through Python/Tensorflow (could have)

Here is the type of thing that may be useful for understanding GANs

[nightrome]

Unfortunately I haven't been working with GANs in a while, so I will be of little use. The project however sounds awesome and I'd be happy to link to it at the top of my page.