[langchain]使用langchain构建第一个llm应用-CSDN博客

本文链接：https://blog.csdn.net/lineuman/article/details/150204441

Build a simple LLM application with chat models and prompt templates

https://python.langchain.com/docs/tutorials/llm_chain/

langchain解决了哪些问题？通过本文的一些演示可以看到，langchain至少解决了1、不同大语言模型的调用方式封装成类似的，例如openai\gemini\claude\deepseeek都是作为一个chatmodal 2、消息格式的转化3、prompt_template

ChatModels的概念

ChatModels are instances of LangChain Runnables, which means they expose a standard interface for interacting with them.
关于langchian 中的Runnable参考https://python.langchain.com/docs/concepts/runnables/

Messages的概念

from langchain_core.messages import SystemMessage, HumanMessage, AIMessage
如果你使用了langchain里面的Message，那么你不需要手动去设置究竟是system\user\assistant\tool了。
在这里插入图片描述

代码演示

在下面的代码演示中，我们可以看到三种invoke的参数，字符串，openai格式的，还有就是langchain_core.messages下的类型


import getpass
import os
from dotenv import load_dotenv
load_dotenv()
from langchain_core.messages import SystemMessage, HumanMessage

from langchain.chat_models import init_chat_model

model = init_chat_model("deepseek-chat", model_provider="deepseek")

messages = [
    SystemMessage(content="Translate the following from English into Chinese."),
    HumanMessage(content="I love programming.")

]

result = model.invoke(messages)
print(result)

print(model.invoke("你是谁?"))

openai_format_messages = [
    {"role": "system", "content": "你是一个信达雅翻译官，把中文翻译成英文"},
    {"role": "user", "content": "这是什么？"},
 
]
print(model.invoke(openai_format_messages))

steam流式调用

import getpass
import os
from dotenv import load_dotenv
load_dotenv()
from langchain_core.messages import SystemMessage, HumanMessage, AIMessage

from langchain.chat_models import init_chat_model

model = init_chat_model("deepseek-chat", model_provider="deepseek")



for token in model.stream("生命的意义?"):
    print(token.content, end="|", flush=True)

Prompt Templates的概念

这里只是简介，想深入了解看：https://python.langchain.com/docs/concepts/prompt_templates/

from langchain_core.prompts import ChatPromptTemplate

import getpass
import os
from dotenv import load_dotenv
load_dotenv()
from langchain_core.prompts import ChatPromptTemplate

from langchain.chat_models import init_chat_model

model = init_chat_model("deepseek-chat", model_provider="deepseek")



prompt_template = ChatPromptTemplate.from_messages(
[("system", "Translate the following from English into {language}, user will provide a sentence in <user_content> and </user_content>, you just need to translate the sentence in <user_content> and </user_content>, and return the translated sentence without any other content."), 
 ("user", "this is user want to translate <user_content> {text}</user_content>")]
)
prompt = prompt_template.invoke({"language": "Janpanse", "text": "I love programming."})

print(prompt.to_messages())

result = model.invoke(prompt.to_messages())
print(result)