web_app/backend/app/answers_model.py

import pinecone
from dotenv import load_dotenv, find_dotenv
from langchain.embeddings.openai import OpenAIEmbeddings
from langchain.memory import ConversationBufferMemory
from langchain.vectorstores import Pinecone
from langchain.chat_models import ChatOpenAI
from langchain.chains import RetrievalQA
from langchain import PromptTemplate

# Load environment variables
load_dotenv(find_dotenv())

# Setting up embeddings
embedder = OpenAIEmbeddings()

# Setting up the vectorstore
pinecone.init(environment="us-west4-gcp")
index = pinecone.Index('seai-website')

#Set up the retriever
vectorstore = Pinecone(index, embedder.embed_query, 'text')

# Setting up the llm
llm = ChatOpenAI(temperature=0.0)

# Setting up the prompt template
template = """
    Use the following context (delimited by <ctx></ctx>) and the chat history (delimited by <hs></hs>) to answer the question:
    ------
    <ctx>
    {context}
    </ctx>
    ------
    <hs>
    {history}
    </hs>
    ------
    {question}
    Answer:
"""
prompt = PromptTemplate(
    input_variables=["history", "context", "question"],
    template=template,
)

# Setting up the memory
memory = ConversationBufferMemory(memory_key="history", input_key="question")

# Setting up the QA chain
qa = RetrievalQA.from_chain_type(
    llm=ChatOpenAI(),
    chain_type='stuff',
    retriever=vectorstore.as_retriever(),
    verbose=True,
    chain_type_kwargs={
        "verbose": True,
        "prompt": prompt,
        "memory": memory
    }
)

def generate_question_response(query: str) -> str:
    return qa.run(query)

if __name__ == "__main__":
    # Sample query
    test_query = "What is the capital of France?"
    
    # Generate a response using the model
    response = generate_question_response(test_query)
    
    # Print the query and its response
    print(f"Query: {test_query}")
    print(f"Response: {response}")