[BUG] - Tutorials should use modern syntax/coding style - torch.rand

[svekars]

Add Link

Multiple links - see below.

Describe the bug

For example: You should use torch.rand(device='cuda') instead of torch.rand(device=torch.device('cuda'))

Instances of this issue:

• amp_recipe.py
• polynomial_autograd.py change to device = "cuda" if torch.cuda.is_available() else "cpu"on L21
• dcgan_faces_tutorial.py
• tuning_guide.py
• nestedtensor.py
• cpp_extension.py
• polynomial_tensor.py
• neural_style_tutorial.py

Describe your environment

• PyTorch 2.0

[malfet]

Another good idea: promote use of torch.set_default_device in tutorials, instead of global device variable, which seems to be a common pattern.

[onurtore]

/assigntome

[Killpit]

/assigntome

[github-actions]

The issue is already assigned. Please pick an opened and unnasigned issue with the docathon-h1-2023 label.

[svekars]

@onurtore are you planing to work on this issue or should we reassign?

[onurtore]

@onurtore are you planing to work on this issue or should we reassign?

Hello,
I am working on it.