Method _build_payload

datastudio/models/openai_api.py:149–195

Build request payload for chat completions API. Args: message: Preprocessed message list (OpenAI format). **kwargs: Override parameters. Returns: Request payload dict.

(self, message: List[dict], **kwargs)

    )

    def _build_payload(self, message: List[dict], **kwargs) -> dict:
        """Build request payload for chat completions API.

        Args:
            message: Preprocessed message list (OpenAI format).
            **kwargs: Override parameters.

        Returns:
            Request payload dict.
        """
        temperature = kwargs.get("temperature", self.temperature)
        max_tokens = kwargs.get("max_tokens", self.max_tokens)
        top_p = kwargs.get("top_p", self.top_p)
        top_k = kwargs.get("top_k", self.top_k)
        min_p = kwargs.get("min_p", self.min_p)
        presence_penalty = kwargs.get("presence_penalty", self.presence_penalty)
        repetition_penalty = kwargs.get("repetition_penalty", self.repetition_penalty)

        # Filter out known keys to avoid duplicates in payload
        extra_kwargs = {
            k: v for k, v in kwargs.items() if k not in self._SAMPLING_PARAM_KEYS
        }

        payload = {
            "model": self.model,
            "messages": message,
            "max_tokens": max_tokens,
            "temperature": temperature,
            "top_p": top_p,
            "top_k": top_k,
            "min_p": min_p,
            "presence_penalty": presence_penalty,
            "repetition_penalty": repetition_penalty,
            **extra_kwargs,
        }

        # Request structured JSON output when return_dict is enabled
        if self.return_dict:
            payload["response_format"] = {"type": "json_object"}

        # Add thinking mode config (for vLLM/SGLang with Qwen3 etc.)
        if "chat_template_kwargs" not in payload:
            payload["chat_template_kwargs"] = {}
        if "enable_thinking" not in payload["chat_template_kwargs"]:
            payload["chat_template_kwargs"]["enable_thinking"] = self.enable_thinking

        return payload

    async def _request_non_streaming(
        self,