fix _load_model

deepeval/experimental/harness/models.py

@@ -1,15 +1,26 @@
-from typing import Optional, Union, List 
+from typing import Optional, Union, List
 
 from deepeval.models.base_model import DeepEvalBaseModel
-from deepeval.experimental.harness.config import GeneralConfig, APIEndpointConfig
+from deepeval.experimental.harness.config import (
+    GeneralConfig,
+    APIEndpointConfig,
+)
 
 
 class DeepEvalHarnessModel(DeepEvalBaseModel):
-    def __init__(self, model_name_or_path: str, model_backend: str, **kwargs) -> None:
-        self.model_name_or_path, self.model_backend = model_name_or_path, model_backend
+    def __init__(
+        self, model_name_or_path: str, model_backend: str, **kwargs
+    ) -> None:
+        self.model_name_or_path, self.model_backend = (
+            model_name_or_path,
+            model_backend,
+        )
         self.additional_params = kwargs
-        super().__init__(model_name=model_name_or_path, **self.additional_params)
+        super().__init__(
+            model_name=model_name_or_path, **self.additional_params
+        )
 
+    def load_model(self, *args, **kwargs):
         try:
             from easy_eval import HarnessEvaluator
         except ImportError as error:
@@ -18,17 +29,26 @@ def __init__(self, model_name_or_path: str, model_backend: str, **kwargs) -> Non
                 "easy_eval is not found."
                 "You can install it using: pip install easy-evaluator"
             )
+
         self.evaluator = HarnessEvaluator(
-            model_name_or_path=self.model_name_or_path, model_backend=self.model_backend, **self.additional_params
+            model_name_or_path=self.model_name_or_path,
+            model_backend=self.model_backend,
+            **self.additional_params,
         )
-
-    def load_model(self, *args, **kwargs):
-        return self.evaluator.llm 
-
-    def _call(self, tasks: List[str], config: Optional[Union[GeneralConfig, APIEndpointConfig]] = None):
-        # TODO: Anthropic is not supported in APIEndpointConfig. 
+        return self.evaluator.llm
+
+    def _call(
+        self,
+        tasks: List[str],
+        config: Optional[Union[GeneralConfig, APIEndpointConfig]] = None,
+    ):
+        # TODO: Anthropic is not supported in APIEndpointConfig.
         if config is None:
-            self.config = APIEndpointConfig() if self.model_name_or_path == "openai" else GeneralConfig()
+            self.config = (
+                APIEndpointConfig()
+                if self.model_name_or_path == "openai"
+                else GeneralConfig()
+            )
         else:
             self.config = config
         return self.evaluator.evaluate(tasks=tasks, config=self.config)