Hi,
I’ve spend the last 2 hours trying to set up pytorch to run with a gpu but unfortunately all I am getting is HTTP 000 connection failures. I have tried in in new environments for py3.6, 3.7 and 3.8. I have also tried it based on both the conda and pip instructions on Get Started Locally and for the instructions listed under previous versions but every time it fails when it hits pytorch, torchvision or torchaudio. All other packages will install until the point including cudatoolkit 10.2. This is specific to PyTorch.
I am installing on a windows 10 enterprise build.
I have also tried other ML libraries with the same cudatoolkit and there is no issue installing those through either conda or pip so I am at a loss as to why these packages specifically fail.
I have previously installed the cpu version and while I ran into the issue for py 3.7, I was able to get around it by installing it for another version.
Any assistance would be appreciated.
Thanks,
dw