Method completion_with_usage

src/harness/llms/client.py:150–159

Make a completion request and return response with usage stats.

(
        self,
        model: str,
        messages: List[Dict[str, Any]],
        **kwargs,
    )

        return litellm.completion(**completion_kwargs)

    def completion_with_usage(
        self,
        model: str,
        messages: List[Dict[str, Any]],
        **kwargs,
    ) -> Tuple[Any, Dict[str, int]]:
        """Make a completion request and return response with usage stats."""
        response = self.completion(model=model, messages=messages, **kwargs)
        usage = extract_token_usage(response.usage)
        return response, usage


_default_client: Optional[LLMClient] = None