Summary
The following code:
from torchmetrics import MeanAbsoluteError, MeanSquaredError, MetricCollection
MetricCollection({
'mse': MeanSquaredError(),
'mae': MeanAbsoluteError(),
})fails in ty 0.0.21 (but not prior versions):
> ty check test.py
error[invalid-argument-type]: Argument to bound method `__init__` is incorrect
--> test.py:3:18
|
1 | from torchmetrics import MeanAbsoluteError, MeanSquaredError, MetricCollection
2 |
3 | MetricCollection({
| __________________^
4 | | 'mse': MeanSquaredError(),
5 | | 'mae': MeanAbsoluteError(),
6 | | })
| |_^ Expected `Metric | MetricCollection | Sequence[Metric | MetricCollection] | dict[str, Metric | MetricCollection]`, found `dict[str, MeanSquaredError | MeanAbsoluteError]`
|
info: Method defined here
--> /Users/Adam/spack/var/spack/environments/default/.spack-env/._view/gd2vtbwe4fojggua4emapinyxnqhryg4/lib/python3.14/site-packages/torchmetrics/collections.py:199:9
|
197 | __jit_unused_properties__: ClassVar[list[str]] = ["metric_state"]
198 |
199 | def __init__(
| ^^^^^^^^
200 | self,
201 | / metrics: Union[
202 | | Metric,
203 | | "MetricCollection",
204 | | Sequence[Union[Metric, "MetricCollection"]],
205 | | dict[str, Union[Metric, "MetricCollection"]],
206 | | ],
| |_________- Parameter declared here
207 | *additional_metrics: Metric,
208 | prefix: Optional[str] = None,
|
info: rule `invalid-argument-type` is enabled by default
Found 1 diagnosticHowever, MeanSquaredError and MeanAbsoluteError are subclasses of Metric, so I believe this should be valid. For now, easiest workaround is to store these in a variable with explicit type hints.
Version
ty 0.0.21
Summary
The following code:
fails in ty 0.0.21 (but not prior versions):
However,
MeanSquaredErrorandMeanAbsoluteErrorare subclasses ofMetric, so I believe this should be valid. For now, easiest workaround is to store these in a variable with explicit type hints.Version
ty 0.0.21