app.py

import gradio as gr
import torch
from transformers import AutoModelForCausalLM, AutoTokenizer
from huggingface_hub import login
import matplotlib.pyplot as plt
import numpy as np

# Login to Hugging Face with token
login(token=os.environ["HF_TOKEN"])

# Liste des modèles
model_list = [
    "meta-llama/Llama-2-13b",
    "meta-llama/Llama-2-7b",
    "meta-llama/Llama-2-70b",
    "meta-llama/Meta-Llama-3-8B",
    "meta-llama/Llama-3.2-3B",
    "meta-llama/Llama-3.1-8B",
    "mistralai/Mistral-7B-v0.1",
    "mistralai/Mixtral-8x7B-v0.1",
    "mistralai/Mistral-7B-v0.3",
    "google/gemma-2-2b",
    "google/gemma-2-9b",
    "google/gemma-2-27b",
    "croissantllm/CroissantLLMBase"
]

# Charger le modèle et le tokenizer
model = None
tokenizer = None

def load_model(model_name):
    global model, tokenizer
    print(f"Chargement du modèle {model_name}...")
    
    tokenizer = AutoTokenizer.from_pretrained(model_name)
    model = AutoModelForCausalLM.from_pretrained(model_name, device_map="auto", torch_dtype=torch.float16, attn_implementation="eager")
    
    print("Modèle chargé avec succès.")
    return f"Modèle {model_name} chargé."

def plot_attention(attention_data):
    tokens = attention_data['tokens']
    attention = attention_data['attention']
    
    fig, ax = plt.subplots(figsize=(10, 10))
    cax = ax.matshow(attention, cmap='viridis')
    fig.colorbar(cax)

    ax.set_xticklabels([''] + tokens, rotation=90)
    ax.set_yticklabels([''] + tokens)
    
    plt.xlabel("Tokens")
    plt.ylabel("Tokens")
    plt.title("Attention Heatmap")
    
    plt.tight_layout()
    plt.savefig('attention_plot.png')
    return 'attention_plot.png'

def plot_probabilities(prob_data):
    words, probs = zip(*prob_data.items())
    
    plt.figure(figsize=(6, 4))
    plt.barh(words, probs, color='skyblue')
    plt.xlabel('Probabilities')
    plt.title('Top Probable Words')
    
    plt.tight_layout()
    plt.savefig('probabilities_plot.png')
    return 'probabilities_plot.png'

def generate_text(input_text, temperature, top_p, top_k):
    global model, tokenizer
    
    inputs = tokenizer(input_text, return_tensors="pt", padding=True, truncation=True, max_length=512).to(model.device)
    
    with torch.no_grad():
        outputs = model.generate(
            **inputs,
            max_new_tokens=50,
            temperature=temperature,
            top_p=top_p,
            top_k=top_k,
            output_scores=True,
            output_attentions=True,
            return_dict_in_generate=True,
            return_legacy_cache=True
        )
    
    generated_text = tokenizer.decode(outputs.sequences[0], skip_special_tokens=True)
    
    # Logits et probabilités du dernier token généré
    last_token_logits = outputs.scores[-1][0]
    probabilities = torch.nn.functional.softmax(last_token_logits, dim=-1)
    
    # Top 5 des mots les plus probables
    top_probs, top_indices = torch.topk(probabilities, 5)
    top_words = [tokenizer.decode([idx.item()]) for idx in top_indices]
    
    prob_data = {word: prob.item() for word, prob in zip(top_words, top_probs)}
    
    # Extraction des attentions
    attentions = torch.cat([att[-1].mean(dim=1) for att in outputs.attentions], dim=0).cpu().numpy()
    attention_data = {
        'attention': attentions,
        'tokens': tokenizer.convert_ids_to_tokens(outputs.sequences[0])
    }
    
    return generated_text, plot_attention(attention_data), plot_probabilities(prob_data)

def reset_app():
    global model, tokenizer
    model = None
    tokenizer = None
    return "Application réinitialisée."

# Interface utilisateur Gradio
with gr.Blocks() as demo:
    with gr.Row():
        model_selection = gr.Accordion("Sélection du modèle", open=True)
        with model_selection:
            model_name = gr.Dropdown(choices=model_list, label="Choisir un modèle", value=model_list[0])
            load_model_button = gr.Button("Charger le modèle")
            load_status = gr.Textbox(label="Statut du modèle", interactive=False)
    
    with gr.Row():
        temperature = gr.Slider(0.0, 1.0, value=0.7, label="Température")
        top_p = gr.Slider(0.0, 1.0, value=0.9, label="Top-p")
        top_k = gr.Slider(1, 100, value=50, label="Top-k")
    
    with gr.Row():
        input_text = gr.Textbox(label="Entrer le texte")
        generate_button = gr.Button("Générer")
    
    with gr.Row():
        output_text = gr.Textbox(label="Texte généré", interactive=False)
    
    with gr.Row():
        attention_plot = gr.Image(label="Carte de chaleur des attentions")
        prob_plot = gr.Image(label="Probabilités des mots les plus probables")
    
    with gr.Row():
        reset_button = gr.Button("Réinitialiser l'application")
    
    load_model_button.click(load_model, inputs=[model_name], outputs=[load_status])
    generate_button.click(generate_text, inputs=[input_text, temperature, top_p, top_k], outputs=[output_text, attention_plot, prob_plot])
    reset_button.click(reset_app)

demo.launch()