Langchain小项目(超级简单，有手就行)

langchain使用的前提准备

1. pip安装

pip install azure-functions azure-storage-blob "qrcode[pil]"

pip install langchain openai streamlit python-dotenv

2. 代码

这里需要在你文件夹内创建一个后缀名为".env"的文件，存储你的Open_AI_Key

from langchain.llms import OpenAI
from dotenv import load_dotenv
import os

load_dotenv('1.env')  # 加载名为 1.env 的文件中的环境变量
OpenAI.api_key = os.getenv('OPENAI_API_KEY')  # 从环境变量中获取 API 密钥

宠物名字小项目

main函数

import langchain_helper as lch
import streamlit as st

st.title("Pet name website")

user_animal_type = st.sidebar.selectbox("What is your pet?", ("Cat", "Dog", "Cow", "Hen",))

if user_animal_type == "Cat":
    language_type = st.sidebar.text_area(label="What language do you want to response?", max_chars=300)

if user_animal_type == "Dog":
    language_type = st.sidebar.text_area(label="What language do you want to response?", max_chars=300)

if user_animal_type == "Cow":
    language_type = st.sidebar.text_area(label="What language do you want to response?", max_chars=300)

if user_animal_type == "Hen":
    language_type = st.sidebar.text_area(label="What language do you want to response?", max_chars=300)

if language_type:
    response = lch.generate_pet_name(user_animal_type, language_type)
    st.text(response['pet_name'])

langchain_helper函数

from langchain.llms import OpenAI
from dotenv import load_dotenv
# 模板
from langchain.prompts import PromptTemplate
# Chain
from langchain.chains import LLMChain
import os
//加载环境变量
load_dotenv()
//加载环境变量
OpenAI.api_key = os.getenv('OPENAI_API_KEY')

//宠物小项目函数
def generate_pet_name(animal_type, language_type):
    llm = OpenAI(temperature=0.5);
	//模板格式
    prompt_template_name = PromptTemplate(
        input_variables=['animal_type', 'language_type'],
        template="I have a {animal_type}，i want you to give me five creative names，and use {language_type} to "
                 "response me."
    )
    //姓名链
    name_chain = LLMChain(llm=llm, prompt=prompt_template_name, output_key="pet_name")
    response = name_chain({'animal_type': animal_type, 'language_type': language_type})

    return response

小结

1. 使用的是Streamlit布置网页,在终端输入streamlit run main.py展示端口网页，从网页中可以选择你的选项来输出想要的结果
2. 关键是def那一部分得搞懂模板格式和姓名链