chore: Resolve all evals mypy errors

PiperOrigin-RevId: 862800470

Author: sararob

vertexai/_genai/_evals_common.py

@@ -71,7 +71,7 @@
 
 
 @contextlib.contextmanager
-def _temp_logger_level(logger_name: str, level: int):
+def _temp_logger_level(logger_name: str, level: int) -> None:
     """Temporarily sets the level of a logger."""
     logger_instance = logging.getLogger(logger_name)
     original_level = logger_instance.getEffectiveLevel()
@@ -95,7 +95,7 @@ def _get_api_client_with_location(
         location,
         api_client.location,
     )
-    return vertexai.Client(
+    return vertexai.Client(  # type: ignore[no-any-return]
         project=api_client.project,
         location=location,
         credentials=api_client._credentials,

vertexai/_genai/_evals_metric_handlers.py

@@ -617,15 +617,15 @@ def _build_pointwise_input(
 
     def _add_autorater_config(self, payload: dict[str, Any]) -> None:
         """Adds autorater config to the request payload if specified."""
-        autorater_config = {}
+        autorater_config: dict[str, Any] = {}
         if self.metric.judge_model:
             autorater_config["autorater_model"] = self.metric.judge_model
         if self.metric.judge_model_generation_config:
             autorater_config["generation_config"] = (
                 self.metric.judge_model_generation_config
             )
         if self.metric.judge_model_sampling_count:
-            autorater_config["sampling_count"] = self.metric.judge_model_sampling_count  # type: ignore[assignment]
+            autorater_config["sampling_count"] = self.metric.judge_model_sampling_count
 
         if not autorater_config:
             return
@@ -989,11 +989,11 @@ def _build_request_payload(
             agent_data=PredefinedMetricHandler._eval_case_to_agent_data(eval_case),
         )
 
-        request_payload = {
+        request_payload: dict[str, Any] = {
             "instance": instance_payload,
         }
 
-        autorater_config = {}
+        autorater_config: dict[str, Any] = {}
         if self.metric.judge_model:
             autorater_config["autorater_model"] = self.metric.judge_model
         if self.metric.judge_model_generation_config:

vertexai/_genai/evals.py

@@ -36,7 +36,7 @@
 try:
     from google.adk.agents import LlmAgent
 except ImportError:
-    LlmAgent = None  # type: ignore[assignment]
+    LlmAgent = None
 
 
 logger = logging.getLogger("vertexai_genai.evals")
@@ -1216,10 +1216,10 @@ def evaluate(
             types.EvaluationDatasetOrDict,
             list[types.EvaluationDatasetOrDict],
         ],
-        metrics: list[types.MetricOrDict] = None,
+        metrics: Optional[list[types.MetricOrDict]] = None,
         location: Optional[str] = None,
         config: Optional[types.EvaluateMethodConfigOrDict] = None,
-        **kwargs,
+        **kwargs: Any,
     ) -> types.EvaluationResult:
         """Evaluates candidate responses in the provided dataset(s) using the specified metrics.
 
@@ -1656,7 +1656,7 @@ def create_evaluation_run(
         evaluation_config = types.EvaluationRunConfig(
             output_config=output_config, metrics=resolved_metrics
         )
-        if agent_info:
+        if agent_info and agent_info.name is not None:
             inference_configs = {}
             inference_configs[agent_info.name] = types.EvaluationRunInferenceConfig(
                 agent_config=types.EvaluationRunAgentConfig(
@@ -2533,7 +2533,7 @@ async def create_evaluation_run(
         if not name:
             name = f"evaluation_run_{uuid.uuid4()}"
 
-        result = await self._create_evaluation_run(  # type: ignore[no-any-return]
+        result = await self._create_evaluation_run(
             name=name,
             display_name=display_name or name,
             data_source=dataset,
@@ -2645,7 +2645,7 @@ async def create_evaluation_item(
         Returns:
           The evaluation item.
         """
-        result = await self._create_evaluation_item(  # type: ignore[no-any-return]
+        result = await self._create_evaluation_item(
             evaluation_item_type=evaluation_item_type,
             gcs_uri=gcs_uri,
             display_name=display_name,
@@ -2676,7 +2676,7 @@ async def create_evaluation_set(
         Returns:
           The evaluation set.
         """
-        result = await self._create_evaluation_set(  # type: ignore[no-any-return]
+        result = await self._create_evaluation_set(
             evaluation_items=evaluation_items,
             display_name=display_name,
             config=config,