diff --git a/tensorflow_decision_forests/component/py_tree/condition.py b/tensorflow_decision_forests/component/py_tree/condition.py
index d672b7d..3aa978c 100644
--- a/tensorflow_decision_forests/component/py_tree/condition.py
+++ b/tensorflow_decision_forests/component/py_tree/condition.py
@@ -301,7 +301,7 @@ def core_condition_to_condition(
                                              core_condition.split_score)
 
   if condition_type.HasField("contains_condition"):
-    items = condition_type.contains_condition.elements
+    items = list(condition_type.contains_condition.elements)
     if not column_spec.categorical.is_already_integerized:
       items = [
           dataspec_lib.categorical_value_idx_to_value(column_spec, item)
diff --git a/tensorflow_decision_forests/keras/keras_test.py b/tensorflow_decision_forests/keras/keras_test.py
index 3d09726..14e5a87 100644
--- a/tensorflow_decision_forests/keras/keras_test.py
+++ b/tensorflow_decision_forests/keras/keras_test.py
@@ -3060,6 +3060,16 @@ def test_abalone_poison_loss(self):
     predictions = model.predict(tf_test)
     logging.info("Predictions: %s", predictions)
 
+  def test_plot_contains_condition(self):
+    df = pd.DataFrame({"f": [-10, 0, 1000], "label": [0, 1, 2]})
+    ds = keras.pd_dataframe_to_tf_dataset(df, label="label")
+    model = keras.CartModel(
+        min_examples=1,
+        features=[keras.FeatureUsage("f", keras.FeatureSemantic.CATEGORICAL)],
+    )
+    model.fit(ds)
+    model_plotter.plot_model(model)
+
 
 if __name__ == "__main__":
   tf.test.main()