feat: GenAI Client(evals) - Add agent data to EvaluationRun show in Vertex AI GenAI SDK evals

PiperOrigin-RevId: 822275512

Author: vertex-sdk-bot

vertexai/_genai/_evals_common.py

@@ -1435,6 +1435,7 @@ def _convert_request_to_dataset_row(
     dict_row["reference"] = request.golden_response
     for candidate in request.candidate_responses:
         dict_row[candidate.candidate] = candidate.text if candidate.text else None
+        dict_row["intermediate_events"] = candidate.events if candidate.events else None
     return dict_row
 
 
@@ -1499,6 +1500,7 @@ def _get_eval_cases_eval_dfs_from_eval_items(
 def _get_eval_result_from_eval_items(
     results: types.EvaluationRunResults,
     eval_items: list[types.EvaluationItem],
+    agent_info: Optional[types.AgentInfo] = None,
 ) -> types.EvaluationResult:
     """Retrieves an EvaluationResult from the EvaluationRunResults.
 
@@ -1522,13 +1524,15 @@ def _get_eval_result_from_eval_items(
         metadata=types.EvaluationRunMetadata(
             candidate_names=candidate_names,
         ),
+        agent_info=agent_info,
     )
     return eval_result
 
 
 def _convert_evaluation_run_results(
     api_client: BaseApiClient,
     evaluation_run_results: types.EvaluationRunResults,
+    agent_info: Optional[types.AgentInfo] = None,
 ) -> list[types.EvaluationItem]:
     """Retrieves an EvaluationItem from the EvaluationRunResults."""
     if not evaluation_run_results or not evaluation_run_results.evaluation_set:
@@ -1545,12 +1549,15 @@ def _convert_evaluation_run_results(
             evals_module.get_evaluation_item(name=item_name)
             for item_name in eval_set.evaluation_items
         ]
-    return _get_eval_result_from_eval_items(evaluation_run_results, eval_items)
+    return _get_eval_result_from_eval_items(
+        evaluation_run_results, eval_items, agent_info
+    )
 
 
 async def _convert_evaluation_run_results_async(
     api_client: BaseApiClient,
     evaluation_run_results: types.EvaluationRunResults,
+    agent_info: Optional[types.AgentInfo] = None,
 ) -> list[types.EvaluationItem]:
     """Retrieves an EvaluationItem from the EvaluationRunResults."""
     if not evaluation_run_results or not evaluation_run_results.evaluation_set:
@@ -1568,7 +1575,9 @@ async def _convert_evaluation_run_results_async(
             for eval_item in eval_set.evaluation_items
         ]
         eval_items = await asyncio.gather(*tasks)
-    return _get_eval_result_from_eval_items(evaluation_run_results, eval_items)
+    return _get_eval_result_from_eval_items(
+        evaluation_run_results, eval_items, agent_info
+    )
 
 
 def _object_to_dict(obj) -> dict[str, Any]:

vertexai/_genai/evals.py

@@ -1293,10 +1293,11 @@ def get_evaluation_run(
         if name.startswith("projects/"):
             name = name.split("/")[-1]
         result = self._get_evaluation_run(name=name, config=config)
+        agent_info = result.agent_info if result.agent_info else None
         if include_evaluation_items:
             result.evaluation_item_results = (
                 _evals_common._convert_evaluation_run_results(
-                    self._api_client, result.evaluation_run_results
+                    self._api_client, result.evaluation_run_results, agent_info
                 )
             )
         return result
@@ -2097,10 +2098,11 @@ async def get_evaluation_run(
         if name.startswith("projects/"):
             name = name.split("/")[-1]
         result = await self._get_evaluation_run(name=name, config=config)
+        agent_info = result.agent_info if result.agent_info else None
         if include_evaluation_items:
             result.evaluation_item_results = (
                 await _evals_common._convert_evaluation_run_results_async(
-                    self._api_client, result.evaluation_run_results
+                    self._api_client, result.evaluation_run_results, agent_info
                 )
             )