When training on 512-size celeba-hq images

[silence1114]

@shepnerd Thank you for your reply. I am very sorry, I made some mistakes in the previous issue. When training on 512-size celeba-hq images. I modified the following parameters in train_options.py：

self.parser.add_argument('--img_shapes', type=str, default='512,512,3',
help='given shape parameters: h,w,c or h,w')
self.parser.add_argument('--mask_shapes', type=str, default='256,256',
help='given mask parameters: h,w')
self.parser.add_argument('--g_cnum', type=int, default=64,
help='# of generator filters in first conv layer')

But it still not work. Are there any other parameters that need to be modified?

[shepnerd]

Thank you for your reminder. First, we need a smaller learning rate (compared with training 256*256 images) like 1e-5 here. Second, using a Gaussian kernel with a large window size and more iterations when computing mask weights, e.g., changing "mask_priority = priority_loss_mask(mask)" to "mask_priority = priority_loss_mask(mask, hsize=128, sigma=1.0 / 60, iters=16)" in network.py (L178). silence14 <notifications@github.com> 于2019年4月2日周二 下午6:02写道：

…

@shepnerd <https://github.com/shepnerd> Thank you for your reply. I am very sorry, I made some mistakes in the previous issue. When training on 512-size celeba-hq images. I modified the following parameters in train_options.py： self.parser.add_argument('--img_shapes', type=str, default='512,512,3', help='given shape parameters: h,w,c or h,w') self.parser.add_argument('--mask_shapes', type=str, default='256,256', help='given mask parameters: h,w') self.parser.add_argument('--g_cnum', type=int, default=64, help='# of generator filters in first conv layer') But it still not work. Are there any other parameters that need to be modified? — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#9>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AE1PCFxTR5jIx0Q-RvU1EddicDlwB00aks5vcyqYgaJpZM4cXon3> .