Fix for runtime error for missing host-to-device copy

ORNL / HydraGNN

Distributed PyTorch implementation of multi-headed graph convolutional neural networks

BSD 3-Clause "New" or "Revised" License

68 stars 29 forks source link

Closed jychoi-hpc closed 5 months ago

jychoi-hpc commented 5 months ago

A quick fix for

RuntimeError: Expected all tensors to be on the same device, but found at least two devices, cuda:0 and cpu!

I found my previous commit has a problem. I missed host-to-device copying. This will be a fix.