Bug: gae.py, matrices on different devices
Howardhuang98 opened this issue · comments
HUANG HAO commented
When I specify the device type = 'gpu' and run the gae.
ga = GAE(input_dim=1, hidden_dim=128, device_type="gpu", epochs=2) ga.learn(x)
Error:
RuntimeError: Expected all tensors to be on the same device, but found at least two devices, cuda:0 and cpu! (when checking argument for argument mat1 in method wrapper_addmm)
I fixed it by modifying gae.py
, line 160: x = torch.from_numpy(data)
to x = torch.from_numpy(data).to(self.device)
Marcus Kalander commented
Hello,
Thank you for the bug report and for taking the time to localize the issue! We will include your fix in the next version of the package.