OpenAI converting API code from GPT-3 to chatGPT-3.5

题意:将OpenAI API代码从GPT-3转换为ChatGPT-3.5

问题背景:

Below is my working code for the GPT-3 API. I am having trouble converting it to work with chatGPT-3.5.

以下是我用于GPT-3 API的工作代码。我在将其转换为适用于ChatGPT-3.5时遇到了困难

php 复制代码
<?php include('../config/config.php'); ?>
<!DOCTYPE html>
<html lang="en">
<head>
<meta charset="UTF-8">
<meta name="viewport" content="width=device-width, initial-scale=1.0">
<meta http-equiv="X-UA-Compatible" content="ie=edge">
<title>Chatbot</title>
<link rel="stylesheet" href="https://cdn.jsdelivr.net/npm/bootstrap-icons@1.10.3/font/bootstrap-icons.css">
<link href="https://cdn.jsdelivr.net/npm/bootstrap@5.3.0-alpha1/dist/css/bootstrap.min.css" rel="stylesheet" integrity="sha384-GLhlTQ8iRABdZLl6O3oVMWSktQOp6b7In1Zl3/Jr59b6EGGoI1aFkw7cmDA6j6gD" crossorigin="anonymous">
<link href="style.css" rel="stylesheet">
</head>
<body>
<div class="container py-5">
  <h1 class="mb-5 text-center">
    <div class="logo"> <img src="/images/Logo-PocketAI.svg" height="80" width="210" aria-label="PocketAI.Online Logo" title="PocketAI.Online Logo" alt="SPocketAI.Online Logo" class="img-fluid"> </div>
  </h1>
  <div class="form-floating mb-3">
    <select class="form-select" id="tab-select" aria-label="Select your purpose">
      <option value="exam" selected>Exam</option>
      <option value="feedback">Feedback</option>
    </select>
    <label for="tab-select">Select your purpose:</label>
  </div>
  <div class="input-group mb-3">
    <div class="form-floating">
      <textarea class="form-control" placeholder="Enter your question or comment here" id="prompt"></textarea>
      <label for="prompt">Enter your question or comment here</label>
    </div>
    <div class="input-group-append username w-100 mt-3 mb-4">
      <button class="btn btn-outline-primary w-100" type="button" id="send-button">Send</button>
    </div>
  </div>
  <div id="output" class="mb-3" style="height: 300px; overflow: auto; border: 1px solid lightgray; padding: 10px;"></div>
  <div id="exam-instructions" class="mb-3" style="display: block;">
    <h3>Exam</h3>
    <p>PocketAI can create multiple choice and true false questions in a format that enables import into Brightspace D2L quizzes using Respondus. Place PocketAI output into a Word document before importing with Respondus. Ask PocketAI questions like the following: <br>
      <br>
      Create 3 multiple choice questions about carbohydrates for a freshman Nutrition online college course.<br>
      Create 2 true false questions about business for a sophomore Business face to face college course.</p>
  </div>
  <div id="feedback-instructions" class="mb-3" style="display: none;">
    <h3>Feedback</h3>
    <p>Enter text to receive writing feedback.</p>
  </div>
</div>
<script>
const previousPrompts = [];
const userName = "<strong>User</strong>";
const chatbotName = "<strong>PocketAI</strong>";

const selectDropdown = document.getElementById("tab-select");

selectDropdown.addEventListener("change", function() {
  const activeTabId = this.value;
  
  // hide all instruction sections
  document.querySelectorAll("[id$='-instructions']").forEach(function(instructionSection) {
    instructionSection.style.display = "none";
  });
  
  // show the instruction section for the active tab
  document.getElementById(`${activeTabId}-instructions`).style.display = "block";
});

document.getElementById("send-button").addEventListener("click", function() {
  const prompt = document.getElementById("prompt").value;
  const activeTabId = selectDropdown.value;

  const endpoint = "https://api.openai.com/v1/completions";
  const apiKey = "<?=$OPEN_AI_KEY;?>";

  document.getElementById("send-button").innerHTML = '<span class="spinner-border spinner-border-sm" role="status" aria-hidden="true"></span> Sending...';

  let promptText = "";
  
  switch (activeTabId) {
    case "exam":
        promptText = "Create quiz questions in the following format: Begin each question with a number followed by a period, and then include the question wording. For each question, include four answer choices listed as letters (A, B, C, D) followed by a period and at least one space before the answer wording. Designate the correct answer by placing an asterisk (*) directly in front of the answer letter (do not put a space between the asterisk and the answer choice). Place the asterisk in front of the answer letter, only the front. It is important that correct answers are identified. Don't make up answers, only select factual answers. For example formatting (don't use this specific example), \"1. What is the recommended daily intake of dietary fiber? A. 10 grams B. 25 grams *C. 50 grams D. 75 grams\". Format true false questions the same way. If you are unsure of the correct answer, don't create the question. Every quiz question and answer must be 100% correct and factual. Do not make up answers. All answers must be correct.";
      break;
     case "feedback":
      promptText = "Can you provide feedback on the writing, grammar, sentence structure, punctuation, and style of this student's paper? The paper should be analyzed for its strengths and weaknesses in terms of written communication. Please provide suggestions for improvement and examples to help the student understand how to make the writing better. The feedback should be specific and provide actionable steps that the student can take to improve their writing skills. Please include at least three examples of areas that could be improved and specific suggestions for how to improve them, such as correcting grammar errors, restructuring sentences, or improving the use of punctuation.";
      break;
  }
  
  const requestData = {
    prompt: previousPrompts.join("\n") + promptText + "\n" + prompt,
    max_tokens: 400,
      model: "text-davinci-003",
    n: 1,
    stop: "",
    temperature: 0.5,
      top_p: 0.0,
      frequency_penalty: 0.0,
      presence_penalty: 0
  };
        
  const requestOptions = {
    method: "POST",
    headers: {
      "Content-Type": "application/json",
      "Authorization": `Bearer ${apiKey}`,
    },
    body: JSON.stringify(requestData),
  };
  
  fetch(endpoint, requestOptions)
    .then(response => response.json())
    .then(data => {
      const reply = data.choices[0].text;
      
      // Add the user message to the chat history
      const userMessage = `<div class="message-container">
        <div class="username">${userName}:&nbsp;</div>
        <div class="user-message">${prompt}</div>
      </div>`;
      document.getElementById("output").innerHTML += userMessage;
      
      const chatbotMessage = `<div class="message-container">
  <div class="username">${chatbotName}:&nbsp;</div>
  <div class="chatbot-message" style="white-space: pre-wrap">${reply}<i class="bi bi-clipboard-check copy-button" data-bs-toggle="tooltip" data-bs-placement="bottom" title="Copy to clipboard" data-text="${reply}" style="cursor: pointer;"></i></div>
</div>`; 
document.getElementById("output").innerHTML += chatbotMessage;

// Add an event listener to each "Copy to Clipboard" button
document.addEventListener("click", function(event) {
  if (event.target.classList.contains("copy-button")) {
    const textToCopy = event.target.dataset.text;
    navigator.clipboard.writeText(textToCopy);
  }
});
     // Scroll to the bottom of the chat history
      document.getElementById("output").scrollTop = document.getElementById("output").scrollHeight;
    
      // Clear the user input field
      document.getElementById("prompt").value = "";
    
      previousPrompts.push(prompt);
      // Clear the spinner and show the "Send" button again
      document.getElementById("send-button").innerHTML = 'Send';
    })
    .catch(error => {
      console.error(error);
    
      // Hide the spinner and show the "Send" button again
      document.getElementById("send-button").innerHTML = 'Send';
    });
});

document.getElementById("prompt").addEventListener("keydown", function(event) {
  if (event.keyCode === 13) {
    event.preventDefault();
    document.getElementById("send-button").
click();
  }
});
</script>
</div>
</div>
<script src="https://cdn.jsdelivr.net/npm/bootstrap@5.3.0-alpha1/dist/js/bootstrap.bundle.min.js" integrity="sha384-w76AqPfDkMBDXo30jS1Sgez6pr3x5MlQ1ZAGC+nuZB+EYdgRZgiwxhTBTkF7CXvN" crossorigin="anonymous"></script>
</body>
</html>

I have read https://openai.com/blog/introducing-chatgpt-and-whisper-apis and referred to this - OpenAI ChatGPT (gpt-3.5-turbo) API: How to access the message content? but still can't make it work.

我已经阅读了https://openai.com/blog/introducing-chatgpt-and-whisper-apis并参考了这个 - OpenAI ChatGPT (gpt-3.5-turbo) API: 如何访问消息内容?但仍然无法使其正常工作

I've tried changing the requestData to this, but no luck:

我尝试将 requestData 更改为如下,但仍然没有成功

const requestData = {
    model: "gpt-3.5-turbo",
    messages: [
      { role: "user", content: prompt }
    ],
    max_tokens: 400,
    temperature: 0.5,
    top_p: 1,
    frequency_penalty: 0,
    presence_penalty: 0
  };

Any help will be greatly appreciated!

任何帮助将不胜感激

问题解决:

better check your requestData object, the GPT 3.5 turbo doesn't need these props

最好检查一下你的 requestData 对象,GPT-3.5 turbo 不需要这些属性

max_tokens,temperature,top_p: 1,frequency_penalty,presence_penalty

I made the same mistake too, GPT 3.5 turbo is way easier to use than I expected. Here's OpenAI sample:

我也犯了同样的错误,GPT-3.5 turbo 比我预期的要简单得多。这里是 OpenAI 的示例

javascript 复制代码
const { Configuration, OpenAIApi } = require("openai");

const configuration = new Configuration({
  apiKey: process.env.OPENAI_API_KEY,
});
const openai = new OpenAIApi(configuration);

const completion = await openai.createChatCompletion({
  model: "gpt-3.5-turbo",
  messages: [{role: "user", content: "Hello world"}],
});
console.log(completion.data.choices[0].message);
相关推荐
踹断瘸子那条好腿.1 分钟前
视觉检测中的深度学习应用
人工智能·深度学习·视觉检测
营赢盈英12 分钟前
OpenAI GPT-3 API: What is the difference between davinci and text-davinci-003?
ai·gpt-3·openai·openai api
程序员阿龙14 分钟前
【2025】基于微信小程序的人工智能课程学习平台的设计与实现(源码+文档+解答)
人工智能·学习·微信小程序·毕业设计·安卓开发·时间管理·在线学习工具
ModelWhale30 分钟前
产教专家共议数字时代下的数据思维人才培养
人工智能·学科建设·高等教育
旗晟机器人1 小时前
焦化行业的变革力量:智能巡检机器人
大数据·人工智能·机器人
2407-2 shw1 小时前
Thinkphp5x远程命令执行 靶场攻略
php·thinkphp
禾高网络1 小时前
护理陪护小程序|陪护系统||陪护系统开发
人工智能·小程序·陪护系统·护工·护理
MavenTalk2 小时前
如何快速上手一个Github的开源项目
人工智能·github·大语言模型·开源项目学习·ai llm
shelly聊AI2 小时前
Github上开源了一款AI虚拟试衣,看看效果
人工智能
量子位2 小时前
姚期智院士大模型新研究:思维图 DoT,用数学理论确保 AI 逻辑一致性
人工智能