Issue running the code with the --generate label

[albusdemens]

Issue running the code with the --generate label

[lucidrains]

@albusdemens I just pushed a new version of the library. From now on, the network settings during training are stored and on generation it will auto instantiate the GAN correctly. You may retroactively fix your problem if you know the image size and network capacity you trained with, and continue training a bit further until the next checkpoint, in which case the problem will correct itself.

[lucidrains]

@albusdemens Hi! When you generate, you have to supply the same network settings as you did when you trained. So the command you would execute would be along the lines of stylegan2_pytorch --image-size 512 --network-capacity {whatever you used} --generate. I'll make this more seamless later today by storing these settings in a dot file

[albusdemens]

Thanks heaps! Everything runs fine now.

…

On Sat, Feb 15, 2020 at 2:32 AM Phil Wang ***@***.***> wrote: @albusdemens <https://github.com/albusdemens> I just pushed a new version of the library. From now on, the network settings during training are stored and on generation it will auto instantiate the GAN correctly. You may retroactively fix your problem if you know the image size and network capacity you trained with, and continue training a bit further until the next checkpoint, in which case the problem will correct itself. — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#15?email_source=notifications&email_token=AACDP25OP5T3LNIE4SMESOLRC5AY7A5CNFSM4KVFPEBKYY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOEL26EWA#issuecomment-586539608>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AACDP27OJRW4GOMFBW3ZKZDRC5AY7ANCNFSM4KVFPEBA> .