The readme writes the cuda version is 10.0 that means the cudatoolkit in pytorch or the cuda version in /usr/local/cuda,i think the build work is by the /usr/local/cuda. i meet the segmentation fault, and not sure the problem comes from cuda version. The cuda version in my path is cuda9.2 and the pytorch environment follow the environment.yaml.
Have you successfully installed pytorch with cuda support (you can check the output of running torch.cuda.is_available())? Could you please post more error messages? Thanks.
The readme writes the cuda version is 10.0 that means the cudatoolkit in pytorch or the cuda version in /usr/local/cuda,i think the build work is by the /usr/local/cuda. i meet the segmentation fault, and not sure the problem comes from cuda version. The cuda version in my path is cuda9.2 and the pytorch environment follow the environment.yaml.