app.py

import os
from langchain_huggingface import HuggingFaceEndpoint
import streamlit as st
from langchain_core.prompts import PromptTemplate
from langchain_core.output_parsers import StrOutputParser

model_id="mistralai/Mistral-7B-Instruct-v0.3"

# Function to get a language model for HuggingFace inference
def get_llm_hf_inference(model_id=model_id, max_new_tokens=128, temperature=0.1):
    """
    Returns a language model for HuggingFace inference.

    Parameters:
    - model_id (str): The ID of the HuggingFace model repository.
    - max_new_tokens (int): The maximum number of new tokens to generate.
    - temperature (float): The temperature for sampling from the model.

    Returns:
    - llm (HuggingFaceEndpoint): The language model for HuggingFace inference.
    """
    llm = HuggingFaceEndpoint(
        repo_id=model_id,
        task="text-generation",
        max_new_tokens=max_new_tokens,
        temperature=temperature,
        token = os.getenv("HF_TOKEN")
    )
    return llm

# Configure the Streamlit app
st.set_page_config(page_title="Chatbot de service après-vente", page_icon="🤗")
st.title("Chatbot de service après-vente")
# A réactiver si on veut une petite description du chatbot
#st.markdown(f"*This is a simple chatbot that uses the HuggingFace transformers library to generate responses to your text input. It uses the {model_id}.*")

# Initialize session state for avatars
if "avatars" not in st.session_state:
    st.session_state.avatars = {'user': None, 'assistant': None}

# Initialize session state for user text input
if 'user_text' not in st.session_state:
    st.session_state.user_text = None

# Initialize session state for model parameters
if "max_response_length" not in st.session_state:
    st.session_state.max_response_length = 256

if "system_message" not in st.session_state:
    st.session_state.system_message = "Tu es un assistant de service après-vente très sympathique qui parlera en français avec un client humain."

if "starter_message" not in st.session_state:
    st.session_state.starter_message = "Bonjour, comment puis-je vous aider aujourd'hui ?"
    
    
# Sidebar for settings
with st.sidebar:
    st.header("Configuration")

    # AI Settings
    st.session_state.system_message = st.text_area(
        "Configuration du contexte de l\'assistant", value="Tu es un assistant de service après-vente très sympathique qui parlera en français avec un client humain."
    )
    st.session_state.starter_message = st.text_area(
        'Premier message de l\'assistant', value="Bonjour, comment puis-je vous aider aujourd'hui ?"
    )

    # Model Settings
    st.session_state.max_response_length = st.number_input(
        "Max Response Length", value=256
    )

    # Avatar Selection
    st.markdown("*Select Avatars:*")
    col1, col2 = st.columns(2)
    with col1:
        st.session_state.avatars['assistant'] = st.selectbox(
            "Avatar de l'assistant", options=["🤗", "💬", "🤖"], index=0
        )
    with col2:
        st.session_state.avatars['user'] = st.selectbox(
            "Avatar de l'utilisateur", options=["👤", "👱‍♂️", "👨🏾", "👩", "👧🏾"], index=0
        )
    # Reset Chat History
    reset_history = st.button("Réinitialiser la conversation")
    
# Initialize or reset chat history
if "chat_history" not in st.session_state or reset_history:
    st.session_state.chat_history = [{"role": "assistant", "content": st.session_state.starter_message}]

def get_response(system_message, chat_history, user_text, 
                 eos_token_id=['User'], max_new_tokens=256, get_llm_hf_kws={}):
    """
    Generates a response from the chatbot model.

    Args:
        system_message (str): The system message for the conversation.
        chat_history (list): The list of previous chat messages.
        user_text (str): The user's input text.
        model_id (str, optional): The ID of the HuggingFace model to use.
        eos_token_id (list, optional): The list of end-of-sentence token IDs.
        max_new_tokens (int, optional): The maximum number of new tokens to generate.
        get_llm_hf_kws (dict, optional): Additional keyword arguments for the get_llm_hf function.

    Returns:
        tuple: A tuple containing the generated response and the updated chat history.
    """
    # Set up the model
    hf = get_llm_hf_inference(max_new_tokens=max_new_tokens, temperature=0.1)

    # Create the prompt template
    prompt = PromptTemplate.from_template(
        (
            "[INST] {system_message}"
            "\nCurrent Conversation:\n{chat_history}\n\n"
            "\nUser: {user_text}.\n [/INST]"
            "\nAI:"
        )
    )
    # Make the chain and bind the prompt
    chat = prompt | hf.bind(skip_prompt=True) | StrOutputParser(output_key='content')

    # Generate the response
    response = chat.invoke(input=dict(system_message=system_message, user_text=user_text, chat_history=chat_history))
    response = response.split("AI:")[-1]

    # Update the chat history
    chat_history.append({'role': 'user', 'content': user_text})
    chat_history.append({'role': 'assistant', 'content': response})
    return response, chat_history

# Chat interface
chat_interface = st.container(border=True)
with chat_interface:
    output_container = st.container()
    st.session_state.user_text = st.chat_input(placeholder="Poser votre question.")
    
# Display chat messages
with output_container:
    # For every message in the history
    for message in st.session_state.chat_history:
        # Skip the system message
        if message['role'] == 'system':
            continue
            
        # Display the chat message using the correct avatar
        with st.chat_message(message['role'], 
                             avatar=st.session_state['avatars'][message['role']]):
            st.markdown(message['content'])
            
 # When the user enter new text:
    if st.session_state.user_text:
        
        # Display the user's new message immediately
        with st.chat_message("user", 
                             avatar=st.session_state.avatars['user']):
            st.markdown(st.session_state.user_text)
            
        # Display a spinner status bar while waiting for the response
        with st.chat_message("assistant", 
                             avatar=st.session_state.avatars['assistant']):

            with st.spinner("Thinking..."):
                # Call the Inference API with the system_prompt, user text, and history
                response, st.session_state.chat_history = get_response(
                    system_message=st.session_state.system_message, 
                    user_text=st.session_state.user_text,
                    chat_history=st.session_state.chat_history,
                    max_new_tokens=st.session_state.max_response_length,
                )
                st.markdown(response)