app.py

import streamlit as st
from langchain_huggingface import HuggingFaceEndpoint
from langchain.prompts import PromptTemplate
from langchain.chains import LLMChain
from langchain.memory import ConversationBufferMemory
import os

# Set up your Hugging Face API token
sec_key = os.getenv('HUGGINGFACE_API_TOKEN')
os.environ['HUGGINGFACE_API_TOKEN'] = sec_key

# Define your Hugging Face model
repo_id = "mistralai/Mixtral-8x7B-Instruct-v0.1"
llm = HuggingFaceEndpoint(repo_id=repo_id, temperature=0.7)

# Define the prompt template
template = """The following is a conversation between a user and an AI assistant.
history:{history}
Final Message by Human: {user_input}
Final Message by AI: """

prompt = PromptTemplate(
    template=template,
    input_variables=["history", "user_input"],
)

# Initialize memory
memory = ConversationBufferMemory()
memory.save_context({"input": "I need some help"}, {"output": "ok! how can i help you."})

# Create the LLM chain
llm_chain = LLMChain(
    prompt=prompt,
    llm=llm,
    memory=memory
)


def generate_response(user_input):
    response = llm_chain.invoke({"history":memory.chat_memory.messages, 'user_input':user_input})
    return response


# Streamlit app
st.title("AI Chatbot")
st.write("Welcome to the AI Chatbot! Ask anything you like.")

# User input
user_input = st.text_input("You:", key="input")

if st.button("Send"):
    if user_input:
        # Generate response
        response = generate_response(user_input)
        response_text = response['text']
        # Display the response
        st.text_area("ChatBot:", response_text, height=100)
        st.write('History:')
        st.write(response['history'])