fix: Fix flaky evaluation test failures caused by thread execution order

PiperOrigin-RevId: 922227627

Author: vertex-sdk-bot

tests/unit/vertexai/test_evaluation.py

@@ -15,6 +15,7 @@
 # limitations under the License.
 #
 
+import json
 import re
 import sys
 import threading
@@ -851,11 +852,26 @@ def test_compute_exact_match_metric(self, api_transport):
         )
         test_metrics = ["exact_match"]
         test_eval_task = EvalTask(dataset=eval_dataset, metrics=test_metrics)
-        mock_metric_results = _MOCK_EXACT_MATCH_RESULT
+
+        def _exact_match_side_effect(**kwargs):
+            request = kwargs.get("request")
+            prediction = request.exact_match_input.instances[0].prediction
+            reference = request.exact_match_input.instances[0].reference
+            score = 1.0 if prediction == reference else 0.0
+            return gapic_evaluation_service_types.EvaluateInstancesResponse(
+                exact_match_results=gapic_evaluation_service_types.ExactMatchResults(
+                    exact_match_metric_values=[
+                        gapic_evaluation_service_types.ExactMatchMetricValue(
+                            score=score
+                        ),
+                    ]
+                )
+            )
+
         with mock.patch.object(
             target=gapic_evaluation_services.EvaluationServiceClient,
             attribute="evaluate_instances",
-            side_effect=mock_metric_results,
+            side_effect=_exact_match_side_effect,
         ):
             test_result = test_eval_task.evaluate()
 
@@ -932,11 +948,26 @@ def test_compute_pointwise_metrics_free_string(self):
             metrics=[_TEST_POINTWISE_METRIC_FREE_STRING],
             metric_column_mapping={"abc": "prompt"},
         )
-        mock_metric_results = _MOCK_POINTWISE_RESULT
+
+        def _pointwise_side_effect(**kwargs):
+            request = kwargs.get("request")
+            instance_data = json.loads(
+                request.pointwise_metric_input.instance.json_instance
+            )
+            # Row with prompt "test_prompt" gets score 5, "text_prompt" gets 4.
+            score = 5 if instance_data.get("abc") == "test_prompt" else 4
+            return gapic_evaluation_service_types.EvaluateInstancesResponse(
+                pointwise_metric_result=(
+                    gapic_evaluation_service_types.PointwiseMetricResult(
+                        score=score, explanation="explanation"
+                    )
+                )
+            )
+
         with mock.patch.object(
             target=gapic_evaluation_services.EvaluationServiceClient,
             attribute="evaluate_instances",
-            side_effect=mock_metric_results,
+            side_effect=_pointwise_side_effect,
         ):
             test_result = test_eval_task.evaluate()
 
@@ -1095,11 +1126,26 @@ def test_compute_pointwise_metrics_without_model_inference(self, api_transport):
         test_eval_task = EvalTask(
             dataset=_TEST_EVAL_DATASET_ALL_INCLUDED, metrics=test_metrics
         )
-        mock_metric_results = _MOCK_SUMMARIZATION_QUALITY_RESULT
+
+        def _summarization_side_effect(**kwargs):
+            request = kwargs.get("request")
+            instance_data = json.loads(
+                request.pointwise_metric_input.instance.json_instance
+            )
+            # Row with response "test" gets score 5, "text" gets score 4.
+            score = 5 if instance_data.get("response") == "test" else 4
+            return gapic_evaluation_service_types.EvaluateInstancesResponse(
+                pointwise_metric_result=(
+                    gapic_evaluation_service_types.PointwiseMetricResult(
+                        score=score, explanation="explanation"
+                    )
+                )
+            )
+
         with mock.patch.object(
             target=gapic_evaluation_services.EvaluationServiceClient,
             attribute="evaluate_instances",
-            side_effect=mock_metric_results,
+            side_effect=_summarization_side_effect,
         ):
             test_result = test_eval_task.evaluate()