on_validation_epoch_end() during sanity checking will not run if val-accuracy-interval is greater than max epochs.

src/deepforest/main.py

@@ -791,9 +791,19 @@ def calculate_empty_frame_accuracy(self, ground_df, predictions_df):
 
     def on_validation_epoch_end(self):
         """Compute metrics."""
+        # Check if we are in a standalone validation run
+        if self.trainer is not None:
+            is_validate_phase = self.trainer.state.fn == "validate"
+        else:
+            is_validate_phase = False
+
+        # Evaluate every n epochs or during standalone validation
+        evaluate_this_epoch = is_validate_phase or (
+            self.config["validation"]["val_accuracy_interval"]
+            <= self.config["train"]["epochs"] and
+            self.current_epoch % self.config["validation"]["val_accuracy_interval"] == 0)
 
-        #Evaluate every n epochs
-        if self.current_epoch % self.config["validation"]["val_accuracy_interval"] == 0:
+        if evaluate_this_epoch:
 
             if len(self.predictions) == 0:
                 return None

tests/test_main.py

@@ -995,3 +995,14 @@ def test_set_labels_invalid_length(m): # Expect a ValueError when setting an inv
     invalid_mapping = {"Object": 0, "Extra": 1}
     with pytest.raises(ValueError):
         m.set_labels(invalid_mapping)
+
+def test_validation_interval_greater_than_epochs(m):
+    # Set interval higher than max_epochs to disable evaluation
+    m.config["validation"]["val_accuracy_interval"] = 3
+    m.config["train"]["epochs"] = 2
+    m.create_trainer()
+    m.trainer.fit(m)
+
+    assert "box_precision" not in m.trainer.logged_metrics
+    assert "box_recall" not in m.trainer.logged_metrics
+    assert "empty_frame_accuracy" not in m.trainer.logged_metrics