Deafult augmentation mode error?

[davidecarnevali]

No description provided.

[davidecarnevali]

The default augmentation mode should be 'ada' (i'm using the stylegan2 pytorch version) but seems to be set always to 'noaug', no matter which --aug value I pass
Indeed if I run:
python train.py -outdir=clusterresults/ --data=/mydata/ --gpus 4 --workers 8 --augpipe bg
I get the following error:
Error: --augpipe cannot be specified with --aug=noaug

The same happens if I try to specify --aug ada meaning the aug variable has always 'noaug' value

Am I missing something or there is an error?

Thank you

[zsyzzsoft]

Please use the --DiffAugment option in this modified version. By default --DiffAugment=color,translation,cutout

[49xxy]

默认增强模式应该是“ada”（我使用的是 stylegan2 pytorch 版本）但似乎总是设置为“noaug”，无论我通过哪个 --aug 值 如果我运行确实： python train.py -outdir=clusterresults/ --data=/mydata/ --gpus 4 --workers 8 --augpipe bg 我收到以下错误： Error: --augpipe cannot be specified with --aug=noaug

如果我尝试指定 --aug ada 意味着 aug 变量始终具有“noaug”值，也会发生同样的情况

我错过了什么还是有错误？

谢谢

Does this mean that Ada and DiffAug cannot be used together？