Update Tensors tutorial — use torch.accelerator for device-agnostic code

[sekyondaMeta]

Description

The Tensors tutorial contains outdated patterns that should be modernized.

Changes needed

Suboptimal / Outdated Patterns

Issue	Current Code	Modern Alternative	Notes
Only checking CUDA ignores other accelerators	if torch.cuda.is_available(): tensor = tensor.to('cuda')	device = torch.accelerator.current_accelerator().type if torch.accelerator.is_available() else 'cpu'	Since PyTorch 2.4+, torch.accelerator provides a device-agnostic API covering CUDA, MPS, XPU, etc.

Files

• beginner_source/blitz/tensor_tutorial.py

[scinqyz]

/assigntome