调用OpenAI的大模型1

木珊数据挖掘

已于 2024-05-28 14:06:08 修改

阅读量1.5k

点赞数 3

分类专栏： LangChain构建大模型应用文章标签： python 人工智能语言模型

于 2024-05-27 17:48:36 首次发布

本文链接：https://blog.csdn.net/m0_45830544/article/details/139243840

版权

LangChain构建大模型应用专栏收录该内容

10 篇文章

订阅专栏

一、通过API调用GPT-3.5

from openai import OpenAI
import os
os.environ['OPENAI_BASE_URL']=''
os.environ['OPENAI_API_KEY']=''

client = OpenAI()
completion = client.chat.completions.create(
  model="gpt-3.5-turbo",
  messages=[
    {"role": "system", "content": "You are a helpful assistant."},
    {"role": "user", "content": "Hello!"}
  ],
    max_tokens=2000,
    temperature=1.8,
    # n=2

)

for item in completion.choices:
    print(item.message.content)

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

# Hello! How can I help you today?

二、流式输出

流式输出简单理解就是模型生成一个词，就输出一个。与之对应的就是模型生成完全部输出后，在一次性全部输出。设置参数stream=true即可实现。

from openai import OpenAI
import os
os.environ['OPENAI_BASE_URL']=''
os.environ['OPENAI_API_KEY']=''

from dotenv import load_dotenv
load_dotenv()

client = OpenAI()
completion = client.chat.completions.create(
  model="gpt-3.5-turbo",
  messages=[
    {"role": "system", "content": "You are a helpful assistant."},
    {"role": "user", "content": "Hello!"}
  ],
    stream=True
)
print(completion)
for item in completion:
    content = item.choices[0].delta.content
    if content is not None:
        print(content,end='')

三、查看token使用数量

调用API需要mon的，有时我们需要查看问题消耗的token数量，定夺是否需要优化提示词，进而控制成本，通过comletion.usage属性查看。

from openai import OpenAI
import os
os.environ['OPENAI_BASE_URL']=''
os.environ['OPENAI_API_KEY']=''

from dotenv import load_dotenv
load_dotenv()

client = OpenAI()
completion = client.chat.completions.create(
  model="gpt-3.5-turbo",
  messages=[
    {"role": "system", "content": "You are a helpful assistant."},
    {"role": "user", "content": "Hello!"}
  ]
)

print(completion.usage)

# CompletionUsage(completion_tokens=9, prompt_tokens=19, total_tokens=28)