app.py

import gradio as gr
from huggingface_hub import InferenceClient
import base64
import io
from PIL import Image
import requests

# Inicialización del cliente de inferencia con el modelo especificado
client = InferenceClient("mistralai/Pixtral-Large-Instruct-2411")

def image_to_base64(image_path):
    """Convert an image file to a base64 string."""
    with open(image_path, "rb") as image_file:
        encoded_string = base64.b64encode(image_file.read()).decode('utf-8')
    return encoded_string

def base64_to_image(base64_string):
    """Convert a base64 string to an image."""
    image_data = base64.b64decode(base64_string)
    image = Image.open(io.BytesIO(image_data))
    return image

def describe_image(image, system_message, max_tokens, temperature, top_p):
    """Describe an image using the model."""
    if image is None:
        return "No image uploaded.", []

    # Convert image to base64
    buffered = io.BytesIO()
    image.save(buffered, format="JPEG")
    image_base64 = base64.b64encode(buffered.getvalue()).decode('utf-8')

    messages = [
        {"role": "system", "content": system_message},
        {"role": "user", "content": "Describe the following image:"},
        {"role": "user", "content": image_base64}
    ]

    response = ""
    for chunk in client.chat_completion(
        messages,
        max_tokens=max_tokens,
        stream=True,
        temperature=temperature,
        top_p=top_p,
    ):
        token = chunk.choices[0].delta.content
        response += token

    return response, [(f"User: Describe the following image:", response)]

def respond(
    user_message: str,
    chat_history: list[tuple[str, str]],
    system_message: str,
    max_tokens: int,
    temperature: float,
    top_p: float,
) -> str:
    """
    Función para generar respuestas basadas en el historial de chat y parámetros de configuración.

    Args:
        user_message (str): Mensaje del usuario.
        chat_history (list[tuple[str, str]]): Historial de chat.
        system_message (str): Mensaje del sistema que define el comportamiento del chatbot.
        max_tokens (int): Máximo número de tokens a generar.
        temperature (float): Temperatura para el muestreo de texto.
        top_p (float): Parámetro top-p para el muestreo de texto.

    Yields:
        str: Respuesta generada por el modelo.
    """
    # Construcción de la lista de mensajes
    messages = [{"role": "system", "content": system_message}]
    for user_msg, assistant_msg in chat_history:
        if user_msg:
            messages.append({"role": "user", "content": user_msg})
        if assistant_msg:
            messages.append({"role": "assistant", "content": assistant_msg})
    messages.append({"role": "user", "content": user_message})

    response = ""

    try:
        # Obtención de la respuesta del modelo
        for chunk in client.chat_completion(
            messages,
            max_tokens=max_tokens,
            stream=True,
            temperature=temperature,
            top_p=top_p,
        ):
            token = chunk.choices[0].delta.content
            response += token
            yield response
    except Exception as e:
        yield f"Error al obtener respuesta: {str(e)}"

def main():
    """
    Función principal para iniciar la interfaz de chat.
    """
    def update_chat(user_message, image, chat_history, system_message, max_tokens, temperature, top_p):
        if image is not None:
            description, new_history = describe_image(image, system_message, max_tokens, temperature, top_p)
            chat_history.extend(new_history)
            user_message = description
        if user_message:
            response_generator = respond(
                user_message,
                chat_history,
                system_message,
                max_tokens,
                temperature,
                top_p,
            )
            for response in response_generator:
                chat_history.append((user_message, response))
                yield "", chat_history, chat_history
        else:
            yield "", chat_history, chat_history

    with gr.Blocks(title="Chatbot con MistralAI", theme=gr.themes.Soft()) as demo:
        gr.Markdown("# Chatbot con MistralAI")
        gr.Markdown("Un chatbot amigable basado en el modelo MistralAI Pixtral-Large-Instruct-2411 que puede describir imágenes y mantener un historial de chat.")

        with gr.Row():
            with gr.Column(scale=3):
                chatbot = gr.Chatbot(label="Conversación")
                user_message = gr.Textbox(label="Mensaje del Usuario", placeholder="Escribe tu mensaje aquí...")
                with gr.Row():
                    submit_button = gr.Button("Enviar")
                    clear_button = gr.Button("Limpiar")

            with gr.Column(scale=2):
                image_input = gr.Image(label="Cargar Imagen", type="pil")
                image_description = gr.Textbox(label="Descripción de la Imagen", interactive=False)

        with gr.Row():
            system_message = gr.Textbox(
                value="You are a friendly Chatbot.",
                label="Mensaje del Sistema",
                placeholder="Define el comportamiento del chatbot."
            )
            max_tokens = gr.Slider(
                minimum=1,
                maximum=2048,
                value=512,
                step=1,
                label="Max New Tokens",
                info="Máximo número de tokens generados."
            )
            temperature = gr.Slider(
                minimum=0.1,
                maximum=4.0,
                value=0.7,
                step=0.1,
                label="Temperature",
                info="Controla la creatividad de la respuesta."
            )
            top_p = gr.Slider(
                minimum=0.1,
                maximum=1.0,
                value=0.95,
                step=0.05,
                label="Top-p (Nucleus Sampling)",
                info="Parámetro para el muestreo del texto."
            )

        chat_history = gr.State([])

        submit_button.click(
            fn=update_chat,
            inputs=[user_message, image_input, chat_history, system_message, max_tokens, temperature, top_p],
            outputs=[user_message, chatbot, chat_history]
        )

        clear_button.click(
            fn=lambda: ([], [], []),
            inputs=[],
            outputs=[user_message, chatbot, chat_history]
        )

        image_input.upload(
            fn=describe_image,
            inputs=[image_input, system_message, max_tokens, temperature, top_p],
            outputs=[image_description, chat_history]
        )

    demo.launch()

if __name__ == "__main__":
    main()