Skip to content

Conversation

@Crissium
Copy link

Running the example command with PyTorch >= 2.6 gives the following error:

[rank0]: Traceback (most recent call last):
[rank0]:   File "/mnt/users_home/cpii.local/yxing/miniconda3/envs/comet/bin/comet-score", line 7, in <module>
[rank0]:     sys.exit(score_command())
[rank0]:              ^^^^^^^^^^^^^^^
[rank0]:   File "/mnt/users_home/cpii.local/yxing/miniconda3/envs/comet/lib/python3.11/site-packages/comet/cli/score.py", line 203, in score_command
[rank0]:     outputs = model.predict(
[rank0]:               ^^^^^^^^^^^^^^
[rank0]:   File "/mnt/users_home/cpii.local/yxing/miniconda3/envs/comet/lib/python3.11/site-packages/comet/models/base.py", line 663, in predict
[rank0]:     predictions = pred_writer.gather_all_predictions()
[rank0]:                   ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
[rank0]:   File "/mnt/users_home/cpii.local/yxing/miniconda3/envs/comet/lib/python3.11/site-packages/comet/models/predict_writer.py", line 99, in gather_all_predictions
[rank0]:     [
[rank0]:   File "/mnt/users_home/cpii.local/yxing/miniconda3/envs/comet/lib/python3.11/site-packages/comet/models/predict_writer.py", line 100, in <listcomp>
[rank0]:     flatten_predictions(torch.load(os.path.join(self.output_dir, f)))
[rank0]:                         ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
[rank0]:   File "/mnt/users_home/cpii.local/yxing/miniconda3/envs/comet/lib/python3.11/site-packages/torch/serialization.py", line 1529, in load
[rank0]:     raise pickle.UnpicklingError(_get_wo_message(str(e))) from None
[rank0]: _pickle.UnpicklingError: Weights only load failed. This file can still be loaded, to do so you have two options, do those steps only if you trust the source of the checkpoint. 
[rank0]:    (1) In PyTorch 2.6, we changed the default value of the `weights_only` argument in `torch.load` from `False` to `True`. Re-running `torch.load` with `weights_only` set to `False` will likely succeed, but it can result in arbitrary code execution. Do it only if you got the file from a trusted source.
[rank0]:    (2) Alternatively, to load with `weights_only=True` please check the recommended steps in the following error message.
[rank0]:    WeightsUnpickler error: Unsupported global: GLOBAL comet.models.utils.Prediction was not an allowed global by default. Please use `torch.serialization.add_safe_globals([comet.models.utils.Prediction])` or the `torch.serialization.safe_globals([comet.models.utils.Prediction])` context manager to allowlist this global if you trust this class/function.

[rank0]: Check the documentation of torch.load to learn more about types accepted by default with weights_only https://pytorch.org/docs/stable/generated/torch.load.html.

Explicitly setting weights_only to False fixed it for me.

@jogonba2
Copy link

For inference, until the PR is accepted, you can monkey-patch torch.load and use the repository as is, without needing to install it in development mode and setting weights_only=False.

# Monkey-patch torch.load for torch>=2.6 compatibility
torch.load = partial(torch.load, weights_only=False)

# Get model output
model_output = model.predict(data, batch_size=batch_size, gpus=gpus)

# Remove monkey-patch for safety
torch.load = partial(torch.load, weights_only=True)

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants