search

junyanz / pytorch-CycleGAN-and-pix2pix

Image-to-Image Translation in PyTorch
Other
23.19k stars 6.33k forks source link

WHETHER USING DROPOUT? #826

Open annihi1ation opened 5 years ago

annihi1ation commented 5 years ago

Hey, professor. When I write a cyclegan myself in pytorch. Then I test it on the orange2apple data set. I find the model totally corrupted. So I wonder if I should clean the dataset or just add some dropout layer in my model?

Looking forward to your reply : )

junyanz commented 5 years ago

Dropout was not used in CycleGAN. But feel free to try it. If you implement your own version, maybe you want to test the code on our datasets first, and try to match the result to our version.