sequential_chain.py

from langchain_huggingface import ChatHuggingFace, HuggingFaceEndpoint
# from langchain.chains import SimpleChain
from langchain_core.prompts import PromptTemplate
from langchain_core.output_parsers import StrOutputParser
from dotenv import load_dotenv

load_dotenv()

# Define the prompt template
prompt1 = PromptTemplate(
    input_variables=["topic"],
    template="Generate a detailed report on {topic}"
)

prompt2 = PromptTemplate(
    input_variables=["text"],
    template="Generate a 5 points summary from the following text: {text}"
)

def init_model():
    llm = HuggingFaceEndpoint(
        repo_id="mistralai/Mixtral-8x22B-Instruct-v0.1",
        # mistralai/Mixtral-8x22B-Instruct-v0.1
        task="text-generation"
    )
    return ChatHuggingFace(llm=llm)


model = init_model()

parser= StrOutputParser()


chain = prompt1 | model | parser | prompt2 | model | parser

result = chain.invoke({"topic": "AI role in Education"})


print(result)  # Should print the French translation of "cricket"

chain.get_graph().print_ascii()