Add type check on maximize in MetricTracker (#1430)

Fixes #1428

CHANGELOG.md

@@ -31,7 +31,7 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 
 ### Fixed
 
--
+- Fixed type checking on the `maximize` parameter at the initialization of `MetricTracker` ([#1428](https://github.com/Lightning-AI/metrics/issues/1428))
 
 
 ## [0.11.0] - 2022-11-30

src/torchmetrics/wrappers/tracker.py

@@ -103,6 +103,8 @@ def __init__(self, metric: Union[Metric, MetricCollection], maximize: Union[bool
             raise ValueError("Argument `maximize` should either be a single bool or list of bool")
         if isinstance(maximize, list) and isinstance(metric, MetricCollection) and len(maximize) != len(metric):
             raise ValueError("The len of argument `maximize` should match the length of the metric collection")
+        if isinstance(metric, Metric) and not isinstance(maximize, bool):
+            raise ValueError("Argument `maximize` should be a single bool when `metric` is a single Metric")
         self.maximize = maximize
 
         self._increment_called = False

tests/unittests/wrappers/test_tracker.py

@@ -41,6 +41,11 @@ def test_raises_error_on_wrong_input():
     ):
         MetricTracker(MetricCollection([MeanAbsoluteError(), MeanSquaredError()]), maximize=[False, False, False])
 
+    with pytest.raises(
+        ValueError, match="Argument `maximize` should be a single bool when `metric` is a single Metric"
+    ):
+        MetricTracker(MeanAbsoluteError(), maximize=[False])
+
 
 @pytest.mark.parametrize(
     "method, method_input",