Handle the error case

src/promptflow-evals/promptflow/evals/evaluate/_code_client/proxy_client.py

@@ -3,6 +3,8 @@
 # ---------------------------------------------------------
 import logging
 
+import numpy as np
+
 from promptflow.client import PFClient
 from promptflow.tracing import ThreadPoolExecutorWithContext as ThreadPoolExecutor
 
@@ -20,9 +22,13 @@ def __init__(self, pf_client: PFClient):
         self._thread_pool = ThreadPoolExecutor(thread_name_prefix="evaluators_thread")
 
     def run(self, flow, data, column_mapping=None, **kwargs):
-        eval_future = self._thread_pool.submit(self._pf_client.run, flow, data=data, column_mapping=column_mapping)
+        eval_future = self._thread_pool.submit(
+            self._pf_client.run, flow, data=data, column_mapping=column_mapping, **kwargs
+        )
         return ProxyRun(run=eval_future)
 
     def get_details(self, proxy_run, all_results=False):
         run = proxy_run.run.result(timeout=60 * 60)
-        return self._pf_client.get_details(run, all_results=all_results)
+        result_df = self._pf_client.get_details(run, all_results=all_results)
+        result_df.replace("(Failed)", np.nan, inplace=True)
+        return result_df

src/promptflow-evals/tests/evals/unittests/test_evaluate.py

@@ -371,7 +371,7 @@ def test_evaluate_output_path(self, evaluate_test_data_jsonl_file, tmpdir, use_t
     def test_evaluate_with_errors(self):
         """Test evaluate_handle_errors"""
         data = _get_file("yeti_questions.jsonl")
-        result = evaluate(data=data, evaluators={"yeti": _yeti_evaluator}, _use_thread_pool=True)
+        result = evaluate(data=data, evaluators={"yeti": _yeti_evaluator})
         result_df = pd.DataFrame(result["rows"])
         expected = pd.read_json(data, lines=True)
         expected.rename(columns={"question": "inputs.question", "answer": "inputs.answer"}, inplace=True)