app.py

import optimum
import transformers
from transformers import AutoConfig, AutoTokenizer, AutoModel, AutoModelForCausalLM
from optimum.bettertransformer import BetterTransformer
import torch
import gradio as gr
import json
import os
import shutil
import requests

# Define the device
device = "cuda" if torch.cuda.is_available() else "cpu"
#Define variables 
temperature=0.4
max_new_tokens=240
top_p=0.92
repetition_penalty=1.7

model_name = "OpenLLM-France/Claire-7B-0.1"

tokenizer = transformers.AutoTokenizer.from_pretrained(model_name)
model = transformers.AutoModelForCausalLM.from_pretrained(model_name,
    device_map="auto",
    torch_dtype=torch.bfloat16,
    load_in_4bit=True                          # For efficient inference, if supported by the GPU card
)
model = BetterTransformer.transform(model)


# Class to encapsulate the Falcon chatbot
class FalconChatBot:
    def __init__(self, system_prompt="Le dialogue suivant est une conversation"):
        self.system_prompt = system_prompt

    def predict(self, user_message, assistant_message, temperature=0.4, max_new_tokens=700, top_p=0.99, repetition_penalty=1.9):
        # Combine the user and assistant messages into a conversation
        conversation = f"{self.system_prompt} {assistant_message if assistant_message else ''} {user_message} "
        # Encode the conversation using the tokenizer
        input_ids = tokenizer.encode(conversation, return_tensors="pt", add_special_tokens=False)
        input_ids = input_ids.to(device)
        # Generate a response using the Falcon model
        response = model.generate(
            input_ids=input_ids, 
            use_cache=False, 
            early_stopping=False, 
            bos_token_id=model.config.bos_token_id, 
            eos_token_id=model.config.eos_token_id, 
            pad_token_id=model.config.eos_token_id, 
            temperature=temperature, 
            do_sample=True, 
            max_new_tokens=max_new_tokens, 
            top_p=top_p, 
            repetition_penalty=repetition_penalty
        )
        # Decode the generated response to text
        response_text = tokenizer.decode(response[0], skip_special_tokens=True)
        return response_text

# Create the Falcon chatbot instance
falcon_bot = FalconChatBot()

# Define the Gradio interface
title = "👋🏻Bienvenue à Tonic's 🌜🌚Claire Chat !"
description = "Vous pouvez utiliser [🌜🌚ClaireGPT](https://huggingface.co/OpenLLM-France/Claire-7B-0.1) Ou dupliquer pour l'uiliser localement ou sur huggingface! [Join me on Discord to build together](https://discord.gg/VqTxc76K3u)."
examples = [
    [
        "Le dialogue suivant est une conversation entre Emmanuel Macron et Elon Musk:",  # user_message
        "[Emmanuel Macron]: Bonjour Monsieur Musk. Je vous remercie de me recevoir aujourd'hui.",  # assistant_message
        150,  # max_new_tokens
        0.9,  # temperature
        0.90,  # top_p
        1.9,  # repetition_penalty
    ]
]

additional_inputs=[
    gr.Textbox("", label="Introduisez Un Personnage Ici ou Mettez En Scene"),
    gr.Slider(
        label="Max new tokens",
        value=100,  # Default value
        minimum=25,
        maximum=256,
        step=1,
        interactive=True,
        info="The maximum numbers of new tokens",
    ),
    gr.Slider(
        label="Temperature",
        value=0.7,  # Default value
        minimum=0.05,
        maximum=1.0,
        step=0.05,
        interactive=True,
        info="Higher values produce more diverse outputs",
    ),
    gr.Slider(
        label="Top-p (nucleus sampling)",
        value=0.90,
        minimum=0.01,
        maximum=0.99,
        step=0.05,
        interactive=True,
        info="Higher values sample more low-probability tokens",
    ),
    gr.Slider(
        label="Repetition penalty",
        value=1.9,
        minimum=1.0,
        maximum=2.0,
        step=0.05,
        interactive=True,
        info="Penalize repeated tokens",
    )
]

iface = gr.Interface(
    fn=falcon_bot.predict,
    title=title,
    description=description,
    examples=examples,
    inputs=[
        gr.Textbox(label="Utilisez se format pour initier une conversation [Personage:]", type="text", lines=5),
    ] + additional_inputs,
    outputs="text",
    theme="ParityError/Anime"
)

# Launch the Gradio interface for the Falcon model
iface.launch()