配置环境

pip install openai

列出模型

import os
import openai
openai.api_key = os.getenv("OPENAI_API_KEY")
openai.Model.list()

发送请求

Create completion

import os
import openai
openai.api_key = ''
openai.Completion.create(
  model="gpt-3.5-turbo",
  prompt="Say this is a test",
  max_tokens=7,
  temperature=0
)

response

{
  "id": "cmpl-uqkvlQyYK7bGYrRHQ0eXlWi7",
  "object": "text_completion",
  "created": 1589478378,
  "model": "text-davinci-003",
  "choices": [
    {
      "text": "\n\nThis is indeed a test",
      "index": 0,
      "logprobs": null,
      "finish_reason": "length"
    }
  ],
  "usage": {
    "prompt_tokens": 5,
    "completion_tokens": 7,
    "total_tokens": 12
  }
}

请求参数

model
prompt: 如果没有, the model will generate as if from the beginning of a new document.
suffix: 感觉没啥用
max_tokens integer Optional Defaults to 16

gpt返回的最大tokens数量, 你的promt的tokens+max_tokens必须小于2048或4096(最新的model)
temperature number Optional Defaults to 1

range from [0,2], 温度越高, 随机性越强

官方建议, 温度和top_p值改变一个
top_p number Optional Defaults to 1 用温度吧, 别用他
n int Opt Defau 1

对于每个prompt, 产生的completions的条数

chat

import os
import openai
openai.api_key = os.getenv("OPENAI_API_KEY")

completion = openai.ChatCompletion.create(
  model="gpt-3.5-turbo",
  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?"}
    ]
)

print(completion.choices[0].message)

response

{
  "id": "chatcmpl-123",
  "object": "chat.completion",
  "created": 1677652288,
  "choices": [{
    "index": 0,
    "message": {
      "role": "assistant",
      "content": "\n\nHello there, how may I assist you today?",
    },
    "finish_reason": "stop"
  }],
  "usage": {
    "prompt_tokens": 9,
    "completion_tokens": 12,
    "total_tokens": 21
  }
}

在 Python 中，可以使用提取助手的回复response[‘choices’][0][‘message’][‘content’]

一般来说，gpt-3.5-turbo-0301对系统消息的关注度不高，因此重要的说明往往放在用户消息中比较好。

edit

给定提示和指令，模型将返回提示的编辑版本。

import os
import openai
openai.api_key = os.getenv("OPENAI_API_KEY")
openai.Edit.create(
  model="text-davinci-edit-001",
  input="What day of the wek is it?",#用作编辑起点的输入文本。
  instruction="Fix the spelling mistakes"# it's like prompt
)

图像生成

建议用midjourney :)

import os
import openai
openai.api_key = os.getenv("OPENAI_API_KEY")
openai.Image.create(
  prompt="A cute baby sea otter",#最长1000个字符
  n=2,#生成的图像数量, 在1~10之间
  size="1024x1024"
)

图像编辑

在给定原始图像和提示的情况下创建编辑或扩展图像。

import os
import openai
openai.api_key = os.getenv("OPENAI_API_KEY")
openai.Image.create_edit(
  image=open("otter.png", "rb"),#要编辑的图像
  mask=open("mask.png", "rb"),#附加图像,可选
  prompt="A cute baby sea otter wearing a beret",#最长1000个字符
  n=2,
  size="1024x1024"
)

嵌入

这个感觉好高级, 留个坑

上传文件

所有文件最大1GB

用到了fine-tunes, 暂时不太懂

一叶小屋

python openai库教程