GroupNorm don't support half precision input

I set net.half(),and groupnorm is float,it got this error:
File “/root/miniconda3/lib/python3.6/site-packages/torch/nn/modules/”, line 493, in call
result = self.forward(*input, **kwargs)
File “/root/miniconda3/lib/python3.6/site-packages/torch/nn/modules/”, line 232, in forward
input, self.num_groups, self.weight, self.bias, self.eps)
File “/root/miniconda3/lib/python3.6/site-packages/torch/nn/”, line 1736, in group_norm
RuntimeError: Expected object of scalar type Float but got scalar type Half for argument #3 ‘tensor1’
but when I use batchnorm,it can run normally.How to fix this question?

Could you post a minimal code snippet to reproduce this issue, please?