[BUG] - advanced_source/semi_structured_sparse.py fails against 2.7

Add Link

advanced_source/semi_structured_sparse.py

Describe the bug

The tutorials are failing with the following error:

Traceback (most recent call last):
  File "/var/lib/workspace/advanced_source/semi_structured_sparse.py", line 451, in <module>
    trainer.train()
  File "/usr/local/lib/python3.10/dist-packages/transformers/trainer.py", line 2241, in train
    return inner_training_loop(
  File "/usr/local/lib/python3.10/dist-packages/transformers/trainer.py", line 2270, in _inner_training_loop
    train_dataloader = self.get_train_dataloader()
  File "/usr/local/lib/python3.10/dist-packages/transformers/trainer.py", line 1029, in get_train_dataloader
    return self.accelerator.prepare(DataLoader(train_dataset, **dataloader_params))
  File "/usr/local/lib/python3.10/dist-packages/accelerate/accelerator.py", line 1398, in prepare
    result = tuple(
  File "/usr/local/lib/python3.10/dist-packages/accelerate/accelerator.py", line 1399, in <genexpr>
    self._prepare_one(obj, first_pass=True, device_placement=d) for obj, d in zip(args, device_placement)
  File "/usr/local/lib/python3.10/dist-packages/accelerate/accelerator.py", line 1270, in _prepare_one
    return self.prepare_data_loader(obj, device_placement=device_placement)
  File "/usr/local/lib/python3.10/dist-packages/accelerate/accelerator.py", line 2222, in prepare_data_loader
    prepared_data_loader = prepare_data_loader(
  File "/usr/local/lib/python3.10/dist-packages/accelerate/data_loader.py", line 1164, in prepare_data_loader
    torch.Generator(device=torch.get_default_device() if hasattr(torch, "get_default_device") else "cpu"),
  File "/var/lib/ci-user/.local/lib/python3.10/site-packages/torch/__init__.py", line 1139, in get_default_device
    device = _GLOBAL_DEVICE_CONTEXT.device_context.device
AttributeError: 'NoneType' object has no attribute 'device'

build log

Please submit fixes against the 2.7-RC-TEST branch and enable in .jenkins/validate_tutorials_built.py.
Possibly related to https://github.com/pytorch/pytorch/issues/131328

Describe your environment

• PyTorch 2.7
• CUDA 12.6

Start with advanced_source/semi_structured_sparse.py and the failing build log, then inspect .jenkins/validate_tutorials_built.py and the related PyTorch issue. Reproduce the tutorial failure against PyTorch 2.7 and determine the compatibility fix; done means the tutorial passes and is enabled in validation on the 2.7-RC-TEST branch.