OpenAI 11.06更新

By Jiamin

OpenAI 11.06更新

在这篇博客中，将讨论 OpenAI 11.06 的一些更新，更新主要有:

  1. 聊天内容支持图片, gpt-4-vision-preview
  2. 返回内容支持json模式
  3. 引入system_fingerprint， 支持可复现性

OpenAI 多模态模型

OpenAI 引入的最令人兴奋的新功能之一是多模态模型，它可以处理文本和图像的组合。这一能力为 AI 应用打开了一个新的维度，从增强的视觉数据分析到更互动的聊天机器人。

GPT-4 Vision: gpt-4-vision-preview

示例: 分析阿里巴巴股票的K线.

alibaba

import openai

openai.api_key = "your-api-key"

response = openai.ChatCompletion.create(
    model="gpt-4-vision-preview",
    messages=[
        {
            "role": "user",
            "content": [
                {"type": "text", "text": "What information can you understand from the K-line of the image?"},
                {
                    "type": "image_url",
                    "image_url": "https://mggg.cloud/img/ali.png",
                },
            ],
        }
    ],
    max_tokens=300,
)

print(response.choices[0].message.content)

output:

The image appears to show a candlestick chart for a stock, specifically ticker 'BABA' which is Alibaba Group Holding Limited. Candlestick charts are commonly used in stock trading to show price movements over time, with each candle representing the trading activity for a specific period.
Each candle shows the opening price, closing price, and price fluctuations within the period it represents. A candle is filled or colored if the closing price is below the opening price (indicating a decrease in price), and it is usually white or hollow if the closing price is above the opening price (indicating an increase in price). The 'wick' or 'shadow' extending from the top or bottom of each candle indicates the high and low prices reached during the period.
Additionally, the chart has moving averages overlaid on it, with different time periods indicated by different colored lines:\n\n- The 5-day moving average (5MA) is the purple line.
- The 10-day moving average (10MA) is the orange line.
- The 20-day moving average (20MA) is the green line.
- The 30-day moving average (30MA) is the red line.
These moving averages smooth out price data by creating a single flowing line and provide insight into the direction of the trend. The chart indicates the moving averages are trending downwards during this period, suggesting that there was a general downtrend in the stock's price. However, near the end of the visible timeline, it appears the trend might be

JSON模式

另一个值得注意的功能是 JSON 模式，它限制模型只生成有效的 JSON 输出。这对于需要将 AI 生成的内容集成到需要结构化数据的应用程序中的开发人员特别有用。

规范输出

假设您需要从给定输入中提取用户的姓名和年龄，并以 JSON 格式返回。有了新的 JSON 模式，这变成了一个简单的任务。

response = openai.ChatCompletion.create(
    model="gpt-3.5-turbo-1106",
    response_format={"type": "json_object"},
    messages=[
        {
            "role": "system",
            "content": [
                {"type": "text", "text": "extract name and age from user input, return a json object"},
            ],
        },
        {
            "role": "user",
            "content": [
                {"type": "text", "text": "my name is wang and my age is 12, "},
            ],
        }
    ],
    max_tokens=300,
)

output = response.choices[0].message.content
import json

user_data = json.loads(output)
print(user_data)

output:

{'name': 'wang', 'age': 12}

System Fingerprint

有唯一id追踪输出

随着 AI 模型不断更新，跟踪可能影响输出的变化的方法非常重要。OpenAI 的系统指纹功能通过为每个模型配置提供唯一标识符来解决这个问题。

response = openai.ChatCompletion.create(
    model="gpt-3.5-turbo-1106",
    response_format={"type": "json_object"},
    messages=[
        {
            "role": "system",
            "content": [
                {"type": "text", "text": "extract name and age from user input, return a json object"},
            ],
        },
        {
            "role": "user",
            "content": [
                {"type": "text", "text": "my name is wang and my age is 12, "},
            ],
        }
    ],
    max_tokens=300,
)

print(response['system_fingerprint'])

output:

'fp_eeff13170a'