Questions about GAN Training

[yzy-thu]

In the official PyTorch Lightning examples GAN training requires manually calling backward and optimizer to ensure the correct optimizer is used at each step.

However, in this repo, it seems optimization is not being done manually. Is this a bug?