@annavaughan This includes your GAN loss version. In the trainer, chose adv_loss_type='mse'. It also includes the option for pure super-resolution in the dataloader by setting pure_sr_ratio=2 or some other integer. Let me know if you have any question on how to use the GANTrainer class. It has a bunch of options but should do most of what you need.
I did not include your version of the generator and discriminator yet. Maybe you could do that yourself, if you think it's necessary, by just adding your version as an option.
@annavaughan This includes your GAN loss version. In the trainer, chose
adv_loss_type='mse'. It also includes the option for pure super-resolution in the dataloader by settingpure_sr_ratio=2or some other integer. Let me know if you have any question on how to use theGANTrainerclass. It has a bunch of options but should do most of what you need.I did not include your version of the generator and discriminator yet. Maybe you could do that yourself, if you think it's necessary, by just adding your version as an option.