why don't you use 2d_cross_entropy but BCE?

pochih / FCN-pytorch

🚘 Easiest Fully Convolutional Networks

406 stars 142 forks source link

Open XavierCHEN34 opened 6 years ago

XavierCHEN34 commented 6 years ago

"nn.BCEWithLogitsLoss()" will this cause overlap between different classes?

pochih commented 6 years ago

Nope.

Water2style commented 5 years ago

哥，你懂了吗，为何做成one-hot然后用BCE呢