keras 2.3.0 breaks this code

[cameronbergh]

causes

` TypeError: gradient_penalty_loss() got an unexpected keyword argument 'sample_weight'

the solution is: pip install 'keras==2.2.5'
`

[ucasiggcas]

I have the same problem,mine is 2.3.1,not compatible

[sxeyyy]

i also have the same issue.

is there a way we can modify the code at all to fix it? i think it might be a similar issue to something described here (tensorflow/addons#1229).

does it relate to an issue (https://stackoverflow.com/questions/57505839/why-am-i-getting-this-unexpected-keyword-argument-typeerror) with the use of *args and **kwargs? in the traceback i do notice *args and **kwargs getting used in tensorflow_backend.py. modifying the code of tensorflow seems pretty daunting, but maybe we can rewrite stylegan.py? how would we go about this?

i was unable to run stylegan.py even after installing keras 2.2.5. that just gave me another error telling me that it didn't support tensorflow 2.x. even after installing tensorflow 1.14 i still got a bunch of errors telling me stuff was depreciated so idk if downgrading can reliably solve things anymore