refactor: adding api_key param to litellm

src/lighteval/metrics/metrics_sample.py

@@ -951,29 +951,28 @@ def __init__(
         short_judge_name: str | None = None,
         response_format: BaseModel | None = None,
         url: str | None = None,
+        api_key: str | None = None,
         hf_provider: str | None = None,
         max_tokens: int | None = None,
         backend_options: dict | None = None,
     ) -> None:
         logger.debug(f"Initializing JudgeLLM with backend: {judge_backend}, model: {judge_model_name}")
 
-        api_key = None
-
         match judge_backend:
             case "openai":
                 if judge_model_name not in self.available_models_openai:
                     raise ValueError(f"{judge_model_name} not in available models for llm as a judge metric")
-                api_key = os.getenv("OPENAI_API_KEY")
+                api_key = api_key or os.getenv("OPENAI_API_KEY")
                 logger.debug("Using OpenAI backend for llm as a judge metric")
 
             case "tgi":
-                api_key = os.getenv("HF_TOKEN")
+                api_key = api_key or os.getenv("HF_TOKEN")
                 if url is None:
                     url = "https://api-inference.huggingface.co/v1/"
                 logger.debug("Using TGI backend")
 
             case "inference-providers":
-                api_key = os.getenv("HF_TOKEN")
+                api_key = api_key or os.getenv("HF_TOKEN")
                 logger.debug("Using Hugging Face Inference backend")
 
             case "litellm":

src/lighteval/metrics/utils/llm_as_judge.py

@@ -329,6 +329,10 @@ def __call_api(prompt):
                     }
                     if max_new_tokens is not None:
                         kwargs["max_tokens"] = (max_new_tokens,)
+                    if self.api_key is not None:
+                        kwargs["api_key"] = self.api_key
+                    if self.url is not None:
+                        kwargs["base_url"] = self.url
 
                     response = litellm.completion(**kwargs)
                     text = response.choices[0].message.content