diff --git a/src/google/adk/evaluation/eval_case.py b/src/google/adk/evaluation/eval_case.py
index 9a787a1..556071c 100644
--- a/src/google/adk/evaluation/eval_case.py
+++ b/src/google/adk/evaluation/eval_case.py
@@ -57,10 +57,10 @@ class Invocation(EvalBaseModel):
   """Content provided by the user in this invocation."""
 
   final_response: Optional[genai_types.Content] = None
-  """Final response from the agent that acts a reference or benchmark."""
+  """Final response from the agent."""
 
   intermediate_data: Optional[IntermediateData] = None
-  """Reference intermediate steps generated as a part of Agent execution.
+  """Intermediate steps generated as a part of Agent execution.
 
   For a multi-agent system, it is also helpful to inspect the route that
   the agent took to generate final response.