torch.nn.modules.module.ModuleAttributeError: 'StaticUnconditionalGAN' object has no attribute '_next_scale_int'

[aloreggia]

Dear authors,
I'm trying to train on an UnconditionalImageDataset using stylegan2_base I got the following error:
torch.nn.modules.module.ModuleAttributeError: 'StaticUnconditionalGAN' object has no attribute '_next_scale_int'

Attached is the config file.
Best,
Andrea

polyp.txt

[LeoXing1996]

Hey @aloreggia, thanks for your interest in our project.
It seems that you use PGGANFetchDataHook in your custom hook. However, this is should only be used for PGGAN or styleGAN1.

[LeoXing1996]

Besides misusing PGGANFetchDataHook, scale in Resize augmentation in the data pipeline should be the same as out_size of the StyleGANv2Generator. In your config, scale and out_size are 256 and 128 respectively.

[aloreggia]

Thank you for your swift reply, now it is working. Sorry, for the silly questions, I'm pretty new with GANs for data augmentation. Do you have any tips for better using your framework for this purpose?

Thanks,
Andrea

[LeoXing1996]

This is not a silly question at all 😄 . You can try to fine-tune a well-trained GAN with your data.