openai.createChatCompletion（） 在 aws lambda 函数中将流转换为 true 时不返回事件

问：

我创建了一个函数，该函数从 返回蒸汽数据。下面的代码片段效果非常好。pythonopenai

def get_data():
            trained_model = train_model(relevent_data, question)

            response = openai.ChatCompletion.create(
                messages=trained_model,
                model=GENERATIVE_MODEL,
                max_tokens=OPENAI_MAX_TOKEN,
                temperature=OPENAI_TEMPERATURE,
                stream=True,
            )

            complete_answer = ""
            for event in response:
                event_text = event["choices"][0]["delta"]
                answer = event_text.get("content", "")  # RETRIEVE CONTENT
                complete_answer += answer
                yield answer

        return Response(get_data(), mimetype="text/event-stream")

由于 Lambda 不支持流式处理，因此使用 ，而是支持 .我切换到 .但是，我面临的问题是，响应不是预期的事件。这是我收到的回复。Pythonnode.jsnode.jsopenai.createChatCompletion

const trained_model = await train_model(relevent_data, question);

const response = await openai.createChatCompletion({
  messages: trained_model,
  model: GENERATIVE_MODEL,
  max_tokens: OPENAI_MAX_TOKEN,
  temperature: OPENAI_TEMPERATURE,
  stream: true,
});

我收到的回复：

 'strict-transport-security': 'max-age=15724800; includeSubDomain      Authorization: 

'Bearer my-openai-key data:    `{"messages":[{"role":"system","content":"My trained model"             Context:"my content"    Answer:"}],"model":"gpt-3.5-turbo","max_tokens":1000,"temperature        'Authorization: Bearer my-openai-key  data: 'data:    {"id":"chatcmpl-7JFj3EkpA02LMyxmiQr7","object":"chat.completion.chunk","created":1684823037,"model":"gpt-3.5-turbo-0301","choices":[{"delta":{"role":"assistant"},"index":0,"finish_    'data:    {"id":"chatcmpl-7JFj3EkpA02LMyxmiQrM6njD4mxh7","object":"chat.completion.chunk","created":1684823037,"model":"gpt-3.5-turbo-0301","choices":[{"delta":{"content":"Hello"},"index":0,"finish_reaso    'data:    {"id":"chatcmpl-7JFj3EkpA02LMyxmiQrM6njD4mxh7","object":"chat.completion.chunk","created":1684823037,"model":"gpt-3.5-turbo-0301","choices":[{"delta":{"content":"!"},"index":0,"finish_reason":n    'data:    {"id":"chatcmpl-7JFj3EkpA02LMyxmiQrM6njD4mxh7","object":"chat.completion.chunk","created":1684823037,"model":"gpt-3.5-turbo-0301","choices":[{"delta":{"content":"    I"},"index":0,"finish_reason":    'data:    {"id":"chatcmpl-7JFj3EkpA02LMyxmiQrM6njD4mxh7","object":"chat.completion.chunk","created":1684823037,"model":"gpt-3.5-turbo-0301","choices":[{"delta":{"content":"    am"},"index":0,"finish_reason"    'data: {"id":"chatcmpl- 'data:    {"id":"chatcmpl-7JFj3EkpA02LMyxmiQrM6njD4mxh7","object":"chat.completion.chunk","created":1684823037,"model":"gpt-3.5-turbo-0301","choices":[{"delta":{"content":"    The"},"index":0,"finish_reason    'data:    {"id":"chatcmpl-7JFj3EkpA02LMyxmiQrM6njD4mxh7","object":"chat.completion.chunk","created":1684823037,"mode'...    1568 more char}

从上面的片段中可以看出，响应是一个对象，如果我使用response.on("data", (chunk) => { console.log("====Cunk=====: ", chunk.toString()); });

它会给我一个错误，响应类型不是事件，而是对象 但是，包含一个字段。 我安慰了其类型，然后我再次尝试，它抛出了一个错误。response.datadeltaresonse.dataStringJSON.parse(response.data)

我很困惑，不确定自己是不是遗漏了什么，我已经向openai createChatCompleteum的官方文档作者求助了

你能帮我为什么会出现这个问题吗？提前感谢您的帮助。

编辑和快速解决方案：

我在这里遇到了一个讨论，我发现这里有一个 npm 包，它目前是有效的。openai-ext

这是我的更新代码：

const trained_model = await train_model(relevent_data, question);
const streamConfig = {
  openai: openai,
  handler: {
    onContent(content, isFinal, stream) {
      console.log("====content===: ", content);

    },
    onDone(stream) {
       console.log("Done!");
    },
    onError(error, stream) {
      console.error(error);
    },
  },
};

await OpenAIExt.streamServerChatCompletion(
  {
    model: GENERATIVE_MODEL,
    messages: trained_model,
    max_tokens: OPENAI_MAX_TOKEN,
    temperature: OPENAI_TEMPERATURE,
  },
  streamConfig
);