向OpenAI API发送含HTML代码的提示词时出现异常结果

我尝试发送包含纯文本和HTML代码的提示词来修正拼写错误，提示词如下：

fix spelling and grammar in the following html code

<p>this is som txt which needs to b fixed.<p>

但通过JavaScript调用OpenAI API提交该提示词时，仅收到如下响应：

而在ChatGPT网页端使用相同提示词可正常得到修复后的结果：

<p>This is some text which needs to be fixed.</p>

附上调用代码，请问哪里操作有误？

const prompt = 'fix spelling and grammar in the following html code\n\n<p>this is som txt which needs to b fixed.<p>';

const response = await fetch(url, {
    method: 'POST',
    headers: {
        'Authorization': bearer,
        'Content-Type': 'application/json'
    },
    body: JSON.stringify({
        temperature: 1,
        top_p: 1,
        n: 1,
        stream: true,
        logprobs: null,
        stop: '\n',
        model: 'gpt-4o',
        messages: [
            { "role": "user", "content": prompt }
        ]
    }),
    signal: controller.signal
});

const reader = response.body.getReader();
const decoder = new TextDecoder("utf-8");

let result = '';

while (true) {
    const { done, value } = await reader.read();
    if (done) {
        break;
    }
    // Massage and parse the chunk of data
    const chunk = decoder.decode(value);
    const lines = chunk.split("\n");
    const parsedLines = lines
        .map((line) => line.replaceAll(/^data: /gi, "").trim()) // Remove the "data: " prefix
        .filter((line) => line !== "" && line !== "[DONE]") // Remove empty lines and "[DONE]"
        .map((line) => JSON.parse(line)); // Parse the JSON string

    for (const parsedLine of parsedLines) {
        const { choices } = parsedLine;
        const { delta } = choices[0];
        const { content } = delta;
        // Update the UI with the new content
        if (content) {
            console.log("Content: " + content);
            result += content;
        }
    }
}
console.log('Result:\n' + result);

错误原因 #

1. stop参数截断了输出：你设置了stop: '\n'，这会强制模型在生成第一个换行符时停止输出。模型返回修复内容时，会先输出```html，紧接着就是换行，刚好触发停止条件，导致后续的HTML代码被截断。
2. temperature参数过高：temperature: 1会让输出随机性极强，不仅可能导致结果不稳定，还可能加剧截断问题，建议调低到0-0.3区间，保证输出精准可控。

修改后的代码 #

const prompt = 'fix spelling and grammar in the following html code\n\n<p>this is som txt which needs to b fixed.<p>';

const response = await fetch(url, {
    method: 'POST',
    headers: {
        'Authorization': bearer,
        'Content-Type': 'application/json'
    },
    body: JSON.stringify({
        temperature: 0.2, // 调低温度保证输出稳定
        top_p: 1,
        n: 1,
        stream: true,
        logprobs: null,
        // 移除stop参数，避免强制截断输出
        model: 'gpt-4o',
        messages: [
            { "role": "user", "content": prompt }
        ]
    }),
    signal: controller.signal
});

// 后续读取流的代码保持不变
const reader = response.body.getReader();
const decoder = new TextDecoder("utf-8");

let result = '';

while (true) {
    const { done, value } = await reader.read();
    if (done) {
        break;
    }
    const chunk = decoder.decode(value);
    const lines = chunk.split("\n");
    const parsedLines = lines
        .map((line) => line.replaceAll(/^data: /gi, "").trim())
        .filter((line) => line !== "" && line !== "[DONE]")
        .map((line) => JSON.parse(line));

    for (const parsedLine of parsedLines) {
        const { choices } = parsedLine;
        const { delta } = choices[0];
        const { content } = delta;
        if (content) {
            console.log("Content: " + content);
            result += content;
        }
    }
}
console.log('Result:\n' + result);

内容的提问来源于stack exchange，提问作者Abid