Haresh

@farshidrayhan this package doesn't support gpu computations.

@farshidrayhan this package doesn't support gpu computations.

delete the ~/.torch/datasets/ade folder and that worked for me.

My network doesn't seem to learn even after 10000 training iterations. the miou is still at 0.20.

equation 5 in the paper says this is epistemic error

This will be useful. thanks !

having the same issue. can someone help ?

Hi, the input image is normalized between 0-1, did you consider using a nn.Sigmoid() activation layer instead of nn.Tanh() as the output activation of your decoder?

I had the same question. I think the output should be sigmoid and not tanh.

That worked ! thanks.