使用 OpenAI 的结构化输出时，示例应如何格式化以实现一次性或几次提示？

Question

使用OpenAI的结构化输出时，one-shot或few-shot提示中的示例在解析之前是否应该遵循结构化输出的JSON格式？

例如，请考虑使用一次性提示的以下代码：

from pydantic import BaseModel
from openai import OpenAI

client = OpenAI()

class CalendarEvent(BaseModel):
    name: str
    date: str
    participants: list[str]

completion = client.beta.chat.completions.parse(
    model="gpt-4o-2024-08-06",
    messages=[
        {"role": "system", "content": "Extract the event information."},
        {"role": "system", "content": "Alice and Bob are going to a science fair on Friday."},
        {"role": "system", "content": '{"name":"Science Fair","date":"Friday","participants":["Alice","Bob"]}'},
        {"role": "user", "content": "Charlie is going to a concert on Saturday night."},
    ],
    response_format=CalendarEvent,
)

event = completion.choices[0].message.parsed

在此示例中，提示“Alice 和 Bob 将在周五参加科学博览会”。伴随着相应的输出：

{"name":"Science Fair","date":"Friday","participants":["Alice","Bob"]}

我的问题是：这些示例是否需要以这样的 JSON 格式构建，或者可以以其他格式呈现？最佳实践是什么？

Answer 1

使用结构化输出时，您通常不需要提供一次性或几次示例。通过将模型作为

response_format=CalendarEvent

传递，您已经将模型的信息包含到最终提示中。

尽管您可以通过向模型及其字段添加描述来改进它，因为它们也包含在最终提示中。

class CalendarEvent(BaseModel):
    """Calender event description"""
    name: str = Field(..., description="The name or title of the event")
    date: str = Field(..., description="The date of the event as a weekday")
    participants: List[str] = Field(..., description="A list of participants' names")

虽然如果您确实想使用一次性或几次提示，则上述 JSON 格式是合适的，但请考虑使用如下标签改进您的系统消息

messages=[
    {"role": "system", "content": "Extract the event information."},
    {"role": "system", "content": """Examples:
                                        User: Alice and Bob are going to a science fair on Friday.
                                        Response: {"name":"Science Fair","date":"Friday","participants":["Alice","Bob"]}"""},
    {"role": "user", "content": "Charlie is going to a concert on Saturday night."},
]