search

microsoft / ProDA

Prototypical Pseudo Label Denoising and Target Structure Learning for Domain Adaptive Semantic Segmentation (CVPR 2021)
https://arxiv.org/abs/2101.10979
MIT License
286 stars 44 forks source link

Training Stage Loss #37

Closed JunkunPeng17 closed 3 years ago

JunkunPeng17 commented 3 years ago

Dear author， Thanks for your great work. Recently I do the code reproduction and fine a small issue which I'm not sure. When I finish the stage1 training and start to train the stage2 I found that the loss will be negative in some cases. After checking the loss code part I thought it will not be negative theoretically. May you know what the problem is or it is not a problem due to my poor knowledge. loss

panzhang0104 commented 3 years ago

The loss includes KLDivLoss which may get a negative loss

JunkunPeng17 commented 3 years ago

Okay, I got it. Thanks for your reply