\n\n翻译结果为：\n\n例行检查\n* 我已确认目前没有类似问题\n* 我已确认我已升级到最新版本\n* 我已完整查看过项目 README,尤其是常见问题部分\n* 我理解并愿意跟进此问题，协助测试和提供反馈\n* 我理解并认可上述内容，并理解项目维护者精力有限，不遵循规则的问题可能会被无视或直接关闭\n\n问题描述\n使用 stream=true 输出时，model_name=chatglm_turbo 无法正常输出，非stream模式可以。\n\n复现步骤\n

import time
from openai import OpenAI
class OneAPIDemo:
    @staticmethod
    def completion(mode_name, messages=[]):
        """
            完成一次对话
        :param mode_name:
        :param messages:
        :return:
        """
        completion = client.chat.completions.create(
            model=model_name,
            messages=messages
        )
        return completion
    @staticmethod
    def completion_stream(model_name, message=[]):
        """
            流式对话
        :param model_name:
        :param message:
        :return:
        """
        completion = client.chat.completions.create(
            model=model_name,
            messages=message,
            stream=True
        )
        return completion
if __name__ == '__main__':
    base_url = ""
    api_key = ""
    client = OpenAI(api_key=api_key,
                    base_url=base_url)
    model_name = "chatglm_turbo"
    messages = [
        {"role": "user", "content": "1+1=?"}
    ]
    start_time = time.time()
    # 是否流式输出
    stream = True
    if stream:
        completion_text = ""
        response = OneAPIDemo.completion_stream(model_name, messages)
        for chunk in response:
            if chunk.choices[0].finish_reason == "stop":
                # 完成
                break
            chunk_text = chunk.choices[0].delta.content
            if chunk_text is not None:
                completion_text += chunk_text
                print(f"接收到: {chunk_text}")
        print(f"回复:{completion_text}\n本次耗时: {time.time() - start_time:.2f}s")
    else:
        response = OneAPIDemo.completion(model_name, messages)
        res_text = response.choices[0].message.content
        print(f"回复: {res_text}\n本次耗时: {time.time() - start_time:.2f}s")

\n\n预期结果\n=2