Fix torch.load() frequently warning in PersistentDataset and GDSDataset

[bnbqq8]

Description

Frequently getting warning massage in a newer Pytorch version (2.4.1 in my case):
"You are using torch.load with weights_only=False (the current default value), which uses the default pickle module implicitly. It is possible to construct malicious pickle data which will execute arbitrary code during unpickling (See https://github.com/pytorch/pytorch/blob/main/SECURITY.md#untrusted-models for more details). In a future release, the default value for weights_only will be flipped to True. This limits the functions that could be executed during unpickling. Arbitrary objects will no longer be allowed to be loaded via this mode unless they are explicitly allowlisted by the user via torch.serialization.add_safe_globals. We recommend you start setting weights_only=True for any use case where you don't have full control of the loaded file. Please open an issue on GitHub for any issues related to this experimental feature."

This pull request fixes an issue with the torch.load() function in the PersistentDataset and GDSDataset classes by adding weights_only=False. The fix ensures that the torch.load() function maintains load consistency in future versions of PyTorch.

Types of changes

• Non-breaking change (fix or new feature that would not break existing functionality).
• Breaking change (fix or new feature that would cause existing functionality to change).
• New tests added to cover the changes.
• Integration tests passed locally by running ./runtests.sh -f -u --net --coverage.
• Quick tests passed locally by running ./runtests.sh --quick --unittests --disttests.
• In-line docstrings updated.
• Documentation updated, tested make html command in the docs/ folder.

[ericspod]

Thanks for the contribution, this is a fix that we do need to remove that warning. I'm not sure if the fails recorded here are related to this PR rather than something else wrong with our CICD system but please run them locally to see if you have the same result.

Could you also try with weights_only=True instead? We should be only loading tensors and not code with the datasets so this should work and be more secure. Thanks!

[bnbqq8]

Thanks for the contribution, this is a fix that we do need to remove that warning. I'm not sure if the fails recorded here are related to this PR rather than something else wrong with our CICD system but please run them locally to see if you have the same result.

Could you also try with weights_only=True instead? We should be only loading tensors and not code with the datasets so this should work and be more secure. Thanks!

Thanks for the comment! I believe the failed checks occurred because the weights_only parameter was not introduced in torch.load() until v1.13.0. This is why the checks for vresion bellow v1.13.0 failed.

So, some compatibility work still needs to be done, and I’m working on it. :)

As for setting weights_only=True, I don’t think it's strictly necessary at this point, since the default setting is weights_only=False. We should maintain consistency for now, right? I mean some users might be caching objects other than tensors, which would require weights_only=False.

[KumoLiu]

Hi @bnbqq8, could you please help address the format issue and DCO?
https://github.com/Project-MONAI/MONAI/actions/runs/11511448642/job/32044911037?pr=8177
https://github.com/Project-MONAI/MONAI/pull/8177/checks?check_run_id=32044901330

[bnbqq8]

@KumoLiu Of course! Thank you for the reminder!

[KumoLiu]

/build