UserWarning at functional confusion_matrix function

[Kr4is]

🐛 Bug

When executing the function 'confusion_matrix' of the functional module, a UserWarning appears:

UserWarning: Torchmetrics v0.9 introduced a new argument class property called `full_state_update` that has
                not been set for this class (_ResultMetric). The property determines if `update` by                                                                                                                                                                                                                            
                default needs access to the full metric state. If this is not the case, significant speedups can be                                                                                                                                                                                                            
                achieved and we recommend setting this to `False`.                                                                                             
                We provide an checking function                                                                                                                
                `from torchmetrics.utilities import check_forward_no_full_state`                                                                               
                that can be used to check if the `full_state_update=True` (old and potential slower behaviour,                                                                                                                                                                                                                 
                default for now) or if `full_state_update=False` can be used safely.

This call does not allow the full_state_update parameter to be passed, so as users we have no way to change this behaviour. So, we should not show a user warning.

To Reproduce

Simply run the confusion_matrix function of the functional module

Code sample

confusion_matrix = torchmetrics.functional.confusion_matrix( preds, targets, num_classes=len(self.classes))

Expected behavior

A UserWarning is not displayed when executing this function.

Environment

• TorchMetrics version (and how you installed TM, e.g. conda, pip, build from source): pip
• Python & PyTorch Version (e.g., 1.0): python3.8, pytorch 1.11
• Any other relevant information such as OS (e.g., Linux): Linux (ubuntu20.04)

[github-actions]

Hi! thanks for your contribution!, great first issue!

[SkafteNicki]

Will be fixed by PR #1058

[tsteffek]

There's also a warning with MeanMetric (and I assume all Aggregators, since full_state_update doesn't appear within the aggregation file). Not entirely sure whether this is related or a separate issue.

[SkafteNicki]

Hi @tsteffek,
It is the same issue since we have not been strict enough to test the new full_state_update feature. It will be added to aggregation metrics in PR #1070 (along some other changes to aggregation metrics).