07. Runnable configuration with @chain decorator

Runnable configuration using @chain decorator

@chain You can add a decorator to transform any function into a chain.

This is a function RunnableLambda Functionally the same as wrapping with.

Let's look at the actual behavior!

Copy

# Configuration file for managing API keys as environment variables
from dotenv import load_dotenv

# Load API key information
load_dotenv()

Copy

 True 

Copy

# LangSmith Set up tracking. https://smith.langchain.com
# !pip install langchain-teddynote
from langchain_teddynote import logging

# Enter a project name.
logging.langsmith("LCEL-Advanced")

Copy

 Start tracking LangSmith. 
[Project name] 
LCEL-Advanced 

Copy

from langchain_core.output_parsers import StrOutputParser
from langchain_core.prompts import ChatPromptTemplate
from langchain_core.runnables import chain
from langchain_openai import ChatOpenAI

ChatPromptTemplate Define two prompt templates using classes.

• prompt1 A short description of the subject given, prompt2 is a request prompt to translate in English.

Copy

# Defines a prompt template.
prompt1 = ChatPromptTemplate.from_template("{topic} Please explain briefly in Korean.")
prompt2 = ChatPromptTemplate.from_template(
    "{sentence} Please make an Instagram post using emoji."
)

custom_chain Functions execute custom chains based on input text.

• @chain Decorate a custom function with a decorator, and the function through decorating Runnable Make it an object.

Copy

@chain
def custom_chain(text):
    # First prompt, ChatOpenAI, Create a chain by concatenating string output parsers.
    chain1 = prompt1 | ChatOpenAI(model="gpt-4o-mini") | StrOutputParser()
    output1 = chain1.invoke({"topic": text})

    # The second prompt, ChatOpenAI, creates a chain by concatenating the string output parser.
    chain2 = prompt2 | ChatOpenAI(model="gpt-4o-mini") | StrOutputParser()
    # Call the second chain, passing it the first parsed result and returning the final result.
    return chain2.invoke({"sentence": output1})

custom_chain is now a viable object runnable ), invoke() I need to run it using

• Check out the LangSmith tracking record, custom_chain There will be tracking information, and below that, OpenAI calls will be nested.

• LangSmith tracking link

Copy

# Call custom_chain
print(custom_chain.invoke("양자역학"))

Copy

 🌌🔬✨ Quantum Mechanics: Explore the mysteries of the micro world!  

Quantum mechanics is a theory that explains the amazing phenomena of atoms and atom particles. 🧪💫 Unlike classical mechanics, it is attractive that the position and momentum of the particles is not known exactly at the same time! 🤔🔍 

Let me introduce some key concepts here:  
-Wave-particle duality 🌊⁇ ️ ⁇ ️ 
-Incorrectity principle ❓🔄 
-Bilateral entanglement 🔗💖 

All of this is the basis of modern physics and technology! 🌍💡 Explore the mysteries of the quantum world together! 🚀🔭 # Quantum mechanics #Physics # The future of science #Mysterious world