Err in 'PyTorch: Tensors' running on GPU

[planewave]

# dtype = torch.device("cuda:0") # Uncomment this to run on GPU
it is supposed to be
# device = torch.device("cuda:0") # Uncomment this to run on GPU
same issue in PyTorch: Tensors and autograd

So x = torch.randn(N, D_in, device=device, dtype=dtype) is the new recommended style to create tensors?

[jxcodetw]

Where did you see this line of code in the tutorial?
Can you post some links?

[planewave]

they are here:

tutorials/beginner_source/examples_tensor/two_layer_net_tensor.py

Line 26 in e55b6e2

# dtype = torch.device("cuda:0") # Uncomment this to run on GPU

tutorials/beginner_source/examples_autograd/two_layer_net_autograd.py

Line 21 in e55b6e2

# dtype = torch.device("cuda:0") # Uncomment this to run on GPU

[gloriouskilka]

@jxcodetw Noticed it here https://pytorch.org/tutorials/beginner/pytorch_with_examples.html