Creating a model on CPU-only pytorch, getting CUDA errors

I get the runtime error “DynamicCUDAInterface::get_device called before CUDA library was loaded” upon trying to call

torch.nn.LSTM(…)

This seems to only happen when I try to do something on a machine without CUDA. However I have the cpu-version of Pytorch installed so I’m not sure why I’m getting this error.

Which version of pytorch are you using?

I am not getting the error in 0.4.1.

For some reason this error happens when you try to pass an array instead of a scalar for (hidden_size = ). I changed that and it was fixed.

I think the error message is odd.