F.binary_cross_entropy_with_logits

torch.nn.functional.binary_cross_entropy_with_logits(input, target, weight=None, size_average=None, reduce=None, reduction='mean', pos_weight=None)

 

Error1: 

RuntimeError: the derivative for 'weight' is not implemented

binary_cross_entropy_with_logits doesn't support back-propagating through the weights attribute.

If you don't need the derivative weights then you can use weights.detach()/weights.data instead of weights.

If you need the derivative, then you'll having to implement binary_cross_entropy_with_logits yourself.

 

Error2: 

RuntimeError: result type Float can't be cast to the desired output type Long

The error points to the target. Convert it to a torch.cuda.FloatTensor using: 

target = target.float()

 

Error3:

RuntimeError: weight tensor should be defined either for all or no classes at /opt/conda/conda-bld/pytorch_1595629408163/work/aten/src/THCUNN/generic/SpatialClassNLLCriterion.cu:27

Check the number of classes and classes_weights.