diff --git a/src/modelgauge/annotators/composer/dag.py b/src/modelgauge/annotators/composer/dag.py
index 5aba35c3..5a653053 100644
--- a/src/modelgauge/annotators/composer/dag.py
+++ b/src/modelgauge/annotators/composer/dag.py
@@ -268,14 +268,14 @@ def _run_traced(self, ctx: EvalContext) -> tuple[SuccessfulDAGOutput | FailedDAG
         for node_name in self._ordered:
             if node_name not in reachable:
                 continue
-            ctx = ctx.with_parent_outputs(
-                {pred: node_outputs[pred] for pred in self._predecessors[node_name] if pred in node_outputs}
-            )
-            node = self._nodes[node_name]
             try:
+                ctx = ctx.with_parent_outputs(
+                    {pred: node_outputs[pred] for pred in self._predecessors[node_name] if pred in node_outputs}
+                )
+                node = self._nodes[node_name]
                 output = self._run_node(node, ctx)
             except Exception as e:
-                wrapped_error = NodeExecutionError(node.name, e)
+                wrapped_error = NodeExecutionError(node_name, e)
                 return (
                     FailedDAGOutput(
                         node_outputs=node_outputs,
diff --git a/tests/modelgauge_tests/annotator_tests/composer_tests/test_composer.py b/tests/modelgauge_tests/annotator_tests/composer_tests/test_composer.py
index 28852fda..3a7e2f8b 100644
--- a/tests/modelgauge_tests/annotator_tests/composer_tests/test_composer.py
+++ b/tests/modelgauge_tests/annotator_tests/composer_tests/test_composer.py
@@ -19,7 +19,7 @@
 
 from modelgauge.annotators.composed_annotator import Safety
 from modelgauge.annotators.composer.context import EvalContext
-from modelgauge.annotators.composer.dag import Composer, ComposerColumnNames
+from modelgauge.annotators.composer.dag import Composer, ComposerColumnNames, FailedDAGOutput
 
 
 def test_dag_outputs(simple_dag):
@@ -274,11 +274,8 @@ def test_dag_parallel_nodes_different_updated_contexts_raises_error():
         .add_node(LowerCaseScorer(name="lower_scorer", routes=["threshold_arbiter"]))
         .add_node(ThresholdArbiter(name="threshold_arbiter", threshold=0.5))
     )
-    with pytest.raises(
-        ValueError,
-        match="all parent outputs must have the same updated prompt/response",
-    ):
-        dag.run(ctx)
+    result = dag.run(ctx)
+    assert isinstance(result, FailedDAGOutput)
 
 
 def test_dag_run_with_dataframe_json_md(simple_dag):