Training
long123524 opened this issue · 2 comments
long123524 commented
Can you provide an example of cldice as a loss function to train the network? I used cldice as a loss function for training, but the result is that I can't train it. The training loss curve is also very strange, and even negative values, and the shock is large, and it cannot be converged.
coreeey commented
I have the same problem :)
Tianyangg commented
From what I understand, please maybe check if your ground truth is one-hot encoded for loss calculation, cz they take foreground channels in 'tsens' calculation and if the ground truth is not one-hot, this term will give zero and result in a very large number.
