Fix to KerasMetricCallback when the model returns unstructured output (#21727)

* Stop doing dict-things to non-dict inputs

* Add a debug check

* Add a debug check

* Remove debug checks, looks good now!

* make fixup

src/transformers/keras_callbacks.py

@@ -231,10 +231,12 @@ class KerasMetricCallback(Callback):
                     # This converts any dict-subclass to a regular dict
                     # Keras REALLY doesn't like it when we pass around a BatchEncoding or other derived class
                     predictions = dict(predictions)
-                if self.output_cols is not None:
-                    predictions = {key: predictions[key] for key in self.output_cols}
-                else:
-                    predictions = {key: val for key, val in predictions.items() if key not in ignore_keys + ["loss"]}
+                    if self.output_cols is not None:
+                        predictions = {key: predictions[key] for key in self.output_cols}
+                    else:
+                        predictions = {
+                            key: val for key, val in predictions.items() if key not in ignore_keys + ["loss"]
+                        }
             prediction_list.append(predictions)
             if not self.use_keras_label:
                 labels = {key: batch[key].numpy() for key in self.label_cols}