[BUG] - FutureWarning in "Introduction to PyTorch - Save and Load the Model" tutorial

[tgilbert09]

Add Link

https://pytorch.org/tutorials/beginner/basics/saveloadrun_tutorial.html

Describe the bug

Getting 2x FutureWarnings from lines model.load_state_dict(torch.load('model_weights.pth')) and model = torch.load('model.pth') in the Save and Load Model tutorial. I think this is what it wants changing:

We recommend you start setting weights_only=True

The full warning I am seeing is:

/var/lib/workspace/beginner_source/basics/saveloadrun_tutorial.py:37: FutureWarning:
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.

Describe your environment

Shows for me in the tutorial on my web browser. Also on MacOS, torch version 2.4.0.

cc @subramen @albanD

[albanD]

cc @mikaylagawarecki to add to our list to update all tutorials with best practices!

[svekars]

I just chatted with @mikaylagawarecki and she gave me a great explanation about the differences. I guess I can send a fix for this one.

[tgilbert09]

Amazing thanks for looking into it so quickly!

If it's useful to know - I spotted another FutureWarning in the TorchVision Object Detection Finetuning Tutorial.

/var/lib/workspace/intermediate_source/engine.py:30: FutureWarning:

torch.cuda.amp.autocast(args...) is deprecated. Please use torch.amp.autocast('cuda', args...) instead.

Coming from https://raw.githubusercontent.com/pytorch/vision/main/references/detection/engine.py. I think others are aware of this one though - pytorch/pytorch#130659 - I will add a comment on that issue :)