search

akanimax / BMSG-GAN

[MSG-GAN] Any body can GAN! Highly stable and robust architecture. Requires little to no hyperparameter tuning. Pytorch Implementation
MIT License
629 stars 105 forks source link

when I train the model ? The d_loss is always 0.0 ? Is there some problems ? #40

Open IwenLeeO opened 4 years ago

IwenLeeO commented 4 years ago

Elapsed [0:05:20.807550] batch: 252 d_loss: 0.000000 g_loss: 9.195433 Elapsed [0:05:32.021947] batch: 261 d_loss: 0.000000 g_loss: 9.544848 Elapsed [0:05:43.313830] batch: 270 d_loss: 0.000000 g_loss: 8.533874 Elapsed [0:05:54.605321] batch: 279 d_loss: 0.000000 g_loss: 8.899170 Elapsed [0:06:06.190011] batch: 288 d_loss: 0.000000 g_loss: 18.478725 Elapsed [0:06:17.612306] batch: 297 d_loss: 0.000000 g_loss: 7.378197

ForawardStar commented 3 years ago

I also have the same issue, have you solved it?

foxy297feng commented 3 years ago

same...

akanimax commented 3 years ago

Hello! The default GAN_loss function being used in this repo is relativistic-hinge. Please refer to this paper for more about this loss function: https://arxiv.org/abs/1807.00734. The discriminator loss for this variant of the GAN_loss remains 0.0 for most of the training, but the GAN trains properly.

foxy297feng commented 3 years ago

Thank you for your reply!

发件人: Animesh Karnewar @.> 发送时间: 2021年4月12日 13:51 收件人: akanimax/BMSG-GAN @.> 抄送: foxy297feng @.>; Comment @.> 主题: Re: [akanimax/BMSG-GAN] when I train the model ? The d_loss is always 0.0 ? Is there some problems ? (#40)

Hello! The default GAN_loss function being used in this repo is relativistic-hinge. Please refer to this paper for more about this loss function: https://arxiv.org/abs/1807.00734. The discriminator loss in for this variant of the GAN_loss remains 0.0 for most of the training, but the GAN trains properly.

― You are receiving this because you commented. Reply to this email directly, view it on GitHubhttps://github.com/akanimax/BMSG-GAN/issues/40#issuecomment-817503421, or unsubscribehttps://github.com/notifications/unsubscribe-auth/ASLDPN6JMIHWQNX57VKX6RLTIKC57ANCNFSM4R4FM3JA.