CrossEntrophy In 1D Data

[herleeyandi]

Hello I still confuse with how cross entrophy loss in pytorch works in 1D data. Here is my condition:
-I have 3 classes
-Input = (NCW) output=(N,W) --> Input(64,3,640), output=(64,640)
-Actually I have tried to use nn.CrossEntrophyLoss but something wrong with dimension, and then I try to unsqueeze it and treat it as image.
-After unsqueeze --> Input = (NCHW) output=(NHW) --> Input(64,3,1,640), output=(64,1,640)
-Then I follow your CrossEntrophyLoss2D function but I got this error:
RuntimeError: invalid argument 3: only batches of spatial targets supported (3D tensors) but got targets of dimension: 2 at /opt/conda/conda-bld/pytorch-nightly_1538995270066/work/aten/src/THNN/generic/SpatialClassNLLCriterion.c:59