loss_semi < = 0.002 - Githubissues

hfslyc / AdvSemiSeg

Adversarial Learning for Semi-supervised Semantic Segmentation, BMVC 2018

502 stars 129 forks source link

loss_semi < = 0.002 #56

Open islzf opened 4 years ago

islzf commented 4 years ago

hi ,what is the problem with my code?what should i do?

tsrobcvai commented 3 years ago

I have the same problem，why？

xiewende commented 2 years ago

the same problems, Did you find a solution?

Holmes2002 commented 1 year ago

I believe that Discriminator causses this problem. Model D is simple so each iter they learn too well . So after running a little time, loss will be 0. U should change training with epoch and about 5 epoch should training D !