[rework] rework the baseline code, tensorflow 1.x to pytorch

[kozistr]

Currently, all of the codes are based on tensorflow 1.x, which is legacy.

so, rework all codes to pytorch code, instead of tensorflow 2.x

[DonaldTsang]

I would actually recommend Tensorflow 2 as it is the most popular, and that it has a distributed network training model that may be useful in some way.

[DonaldTsang]

Just checked the popularity of Tensorflow 2 over PyTorch, and they are on par with one another...
Are there any Keras-like libraries that support both TF and PyTorch?
(I guess the game is on between Google's TF and Facebook's PyTorch)

[kozistr]

Just checked the popularity of Tensorflow 2 over PyTorch, and they are on par with one another...
Are there any Keras-like libraries that support both TF and PyTorch?
(I guess the game is on between Google's TF and Facebook's PyTorch)

maybe, there's no library that supports both of them as i know.

I also think TF2 is popular than pytorch 1.x.
But, in the aspect of developing the GANs, i also think pure keras framework is not suitable and tricky to implement for controlling the low-level stuffs than pytorch. So, now i'm considering between them : ).

[DonaldTsang]

So I would make that claim that whatever is more popular and easier to implement the underlying features should be used, that would be Tensorflow in this case.