completions

Migrate to the Responses API

https://platform.openai.com/docs/guides/migrate-to-responses

Chat Completions API

from openai import OpenAI
client = OpenAI()

completion = client.chat.completions.create(
  model="gpt-5",
  messages=[
      {
          "role": "user",
          "content": "Write a one-sentence bedtime story about a unicorn."
      }
  ]
)

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

Responses API

from openai import OpenAI
client = OpenAI()

response = client.responses.create(
  model="gpt-5",
  input="Write a one-sentence bedtime story about a unicorn."
)

print(response.output_text


https://www.devze.com/dev/1748325Inq6YxO0o

• 总结

 

chat/completions 和 completions 是 OpenAI API 中的两个不同的端点，它们提供了不同的功能和交互模式。以下是它们的主要区别：

 

completions 端点

用途:

• 主要用于生成文本补全。你提android供一个提示（prompt），模型会基于这个提示生成后续的文本。

交互模式:

• 单次请求-响应模式。你发送一个提示，模型返回一个补全结果。

适用场景:

• 适用于需要连续生成文本的场景，如编写文章、代码补全、生成故事等。

示例请求:

{
    "model": "text-davinci-003",
    "prompt": "Once upon a time, in a land far, far away,",
    "max_tokens": 100
}

示例响应:

{
    "id": "cmpl-5eU3oZz1w9Q8Jt3B3o5Q5Z5Z1",
    "object": "text_completion",
    "created": 1609459200,
    "model": js"text-davinci-003",
    "choices": [
        {
            "text": " there lived a wise old owl who knew all the secrets of the forest...",
            "index": 0,
            "logprobs": null,
            "finish_reason": "length"
        }
    ],
    "usage": {
        "prompt_tokens": 10,
        "completion_tokens": 100,
        "total_tokens": 110
    }
}

 

chat/completions 端点

用途:

• 主要用于对话生成。你提供一系列对话消息，模型会基于这些消息生成下一条回复。

交互模式:

• 多轮对话模式。你可以提供一个包含多轮对话的消息列表，模型会基于整个对话上下文生成回复。

适用场景:

• 适用于需要多轮对话的场景，如聊天机器人、客户服务、对话系统等。

示例请求:

{
    "model": "gpt-4",
    "messages": [
        {"role": "system", "content": "You are a helpful assistant."},
        {"role": "user", "content": "Who won the world series in 2020?"},
        {"role": "assistant", "content": "The Los Angeles Dodgers won the World Series in 2020."},
        {"role": "user", "content": "Where was it played?"}
    ]
}

示例响应:

{
    "id": "chatcmpl-5eU3oZz1w9Q8Jt3B3编程o5Q5Z5Z1",
    "object": "chat.completion",
    "created": 1609459200,
    "model": "gpt-4",
    "choices": [
        {
            "message": {
                "role": "assandroidistant",
                "content": "The 2020 World Series was played at Globe Life Field in Arlington, Texas."
            },
            "index": 0,
            "finish_reason": "stop"
        }
    ],
    "usage": {
        "prompt_tokens": 50,
        "completion_tokens": 20,
        "total_tokens": 70
    }
}

 

总结

• completions 端点适用于单次文本补全任务，通常用于连续文本生成。
• chat/completions 端点适用于多轮对话生成任务，提供更自然的对话体验。

选择哪个端点取决于你的具体需求。

• 如果你需要生成连续的文本，completions 端点可能更合适。
• 如果你需要处理多轮对话，chat/completions 端点会更适合。

 

https://github.com/datawhalechina/hello-agents/blob/main/code/chapter11/01_dataset_loading.py

def load_sft_dataset():
    """
    使用RLTrainingTool加载SFT格式的GSM8K数据集

    SFT数据格式:
    {
        "prompt": "Question: ...\n\nLet's solve this step by step:\n",
        "completion": "Step 1: ...\nFinal Answer: 42",
        "text": "Question: ...\n\nLet's solve this step by step:\nStep 1: ...\nFinal Answer: 42"
    }