Transition to tensorflow.compat.v1 as tf, instead of tensorflow as tf about hypergan

[IIIBlueberry]

This seems like a doable short term fix, yes, but I am uncertain how long that compatibility would be supported. It is after all harder to downgrade.

AFIK the compatibility modules is here to stay, tensorflow 1.15 also included the TF 2.0 API which allowed you to write forward compatible code

TensorFlow 1.15 contains a complete implementation of the 2.0 API in its compat.v2 module. It contains a copy of the 1.15 main module (without contrib) in the compat.v1 module. TensorFlow 1.15 is able to emulate 2.0 behavior using the enable_v2_behavior() function.
This enables writing forward compatible code: by explicitly importing either tensorflow.compat.v1 or tensorflow.compat.v2, you can ensure that your code works without modifications against an installation of 1.15 or 2.0.

However with the hypergan develop branch it seems that hypergan author is transitioning to pytorch #259

[LivelyCarpet87]

This seems like a doable short term fix, yes, but I am uncertain how long that compatibility would be supported. It is after all harder to downgrade.