Currently you can only load pre-trained Generator or Discriminator from pretrained_gen_path and pretrained_dis_path respectively (refer to config.py).
The simplest way to load pre-trained Generator from save/**/models is to directly replace the state dict in pretrained_gen_path with the state dict in save/**/models.
If you want to load the checkpoint of adversarial training, you can easily update the code of the function self._save() in instructor.py with the code of saving Discriminator. Therefore, you can regard the state dict of model saved during adversarial training as the state dict of model saved after the pre-training.
Currently you can only load pre-trained Generator or Discriminator from
pretrained_gen_pathandpretrained_dis_pathrespectively (refer toconfig.py). The simplest way to load pre-trained Generator fromsave/**/modelsis to directly replace the state dict inpretrained_gen_pathwith the state dict insave/**/models. If you want to load the checkpoint of adversarial training, you can easily update the code of the functionself._save()ininstructor.pywith the code of saving Discriminator. Therefore, you can regard the state dict of model saved during adversarial training as the state dict of model saved after the pre-training.