move lr scheduler config

clinicadl/maps_manager/maps_manager.py

@@ -445,7 +445,8 @@ def _check_args(self, parameters):
             self.n_classes = output_size(self.network_task, None, train_df, self.label)
         else:
             self.n_classes = None
-
+        print(self.network_task)
+        print(evaluation_metrics(network_task=self.network_task))
         self.metrics_module = MetricModule(
             evaluation_metrics(network_task=self.network_task),
             n_classes=self.n_classes,

clinicadl/metrics/metric_module.py

@@ -33,6 +33,7 @@ def __init__(self, metrics, n_classes=2):
             for method_name in dir(MetricModule)
             if callable(getattr(MetricModule, method_name))
         ]
+        print(metrics)
         self.metrics = dict()
         for metric in metrics:
             if f"compute_{metric.lower()}" in list_fn:

clinicadl/trainer/tasks_utils.py

@@ -192,11 +192,19 @@ def evaluation_metrics(network_task: Union[str, Task]):
     """
     network_task = Task(network_task)
     if network_task == Task.CLASSIFICATION:
-        return [e.value for e in ClassificationMetric].remove("loss")
+        x = [e.value for e in ClassificationMetric]
+        x.remove("loss")
+        return x
     elif network_task == Task.REGRESSION:
-        return [e.value for e in RegressionMetric].remove("loss")
+        x = [e.value for e in RegressionMetric]
+        x.remove("loss")
+        return x
     elif network_task == Task.RECONSTRUCTION:
-        return [e.value for e in ReconstructionMetric].remove("loss")
+        x = [e.value for e in ReconstructionMetric]
+        x.remove("loss")
+        return x
+    else:
+        raise ValueError("Unknown network task")
 
 
 def test(