调用OpenAI GPT-3 API遇请求超时问题求助

调用OpenAI GPT-3 API时持续出现超时错误：

Request timed out: HTTPSConnectionPool(host='api.openai.com', port=443): Read timed out. (read timeout=600)

相关代码如下：

import time
import openai as ai

def generate_gpt3_response(user_text, print_output=False):
    """
    Query OpenAI GPT-3 for the specific key and get back a response
    :type user_text: str the user's text to query for
    :type print_output: boolean whether or not to print the raw output JSON
    """
    time.sleep(5)
    completions = ai.Completion.create(
        engine='text-davinci-003',  # Determines the quality, speed, and cost.
        temperature=0.5,            # Level of creativity in the response
        prompt=user_text,           # What the user typed in
        max_tokens=150,             # Maximum tokens in the prompt AND response
        n=1,                        # The number of completions to generate
        stop=None,                  # An optional setting to control response generation
    )

    # Displaying the output can be helpful if things go wrong
    if print_output:
        print(completions)

    # Return the first choice's text
    return completions.choices[0].text

df_test['GPT'] = df_test['Q20'].apply(lambda x: \
              generate_gpt3_response\
              ("I am giving you the answer of respondents \
                in the format [Q20], \
                give me the Broader topics like customer service, technology, satisfaction\
                or the related high level topics in one word in the \
                format[Topic: your primary topic] for the text '{}' ".format(x)))

# 提取结果
df_test['GPT'] = df_test['GPT'].apply(lambda x: (x.split(':')[1]).replace(']',''))

已尝试修改参数但错误仍未解决，求可行解决方案。

可行解决方案 #

• 添加请求重试机制：网络波动或API临时限流是超时常见原因，用重试机制自动重试失败请求，避免单次中断：

  from tenacity import retry, stop_after_attempt, wait_exponential, retry_if_exception_type
  import requests
  
  @retry(stop=stop_after_attempt(3),
        wait=wait_exponential(multiplier=1, min=2, max=10),
        retry=retry_if_exception_type((requests.exceptions.Timeout, requests.exceptions.ConnectionError)))
  def generate_gpt3_response(user_text, print_output=False):
      time.sleep(5)
      completions = ai.Completion.create(
          engine='text-davinci-003',
          temperature=0.5,
          prompt=user_text,
          max_tokens=150,
          n=1,
          stop=None,
          timeout=300  # 缩短超时阈值，避免无效等待
      )
      if print_output:
          print(completions)
      return completions.choices[0].text
  

• 调整超时参数：当前read timeout=600过长，API正常响应无需这么久，建议在ai.Completion.create()中显式设置timeout=300（或更短的120），覆盖默认超时配置。

• 优化请求方式：用apply逐行调用API会增加连接开销，改用循环遍历并记录失败项，同时控制请求频率：

  from tqdm import tqdm
  
  results = []
  for text in tqdm(df_test['Q20'].tolist()):
      try:
          res = generate_gpt3_response(text)
          results.append(res)
      except Exception as e:
          results.append(None)  # 失败项标记后统一处理
  df_test['GPT'] = results
  

• 检查网络环境：确认本地网络可稳定访问api.openai.com，尝试切换网络（如手机热点）或排查代理、防火墙是否拦截HTTPS连接。

• 更换模型引擎：text-davinci-003属于旧版模型，响应稳定性不如新版gpt-3.5-turbo-instruct，替换后可提升响应速度并降低成本：

  completions = ai.Completion.create(
      engine='gpt-3.5-turbo-instruct',  # 替换为新版指令模型
      temperature=0.5,
      prompt=user_text,
      max_tokens=150,
      n=1,
      stop=None,
      timeout=300
  )
  

内容的提问来源于stack exchange，提问作者opsv