update

app.py

@@ -1,59 +1,23 @@
-# Fichier app.py
-import gradio as gr
 from transformers import AutoModelForCausalLM, AutoTokenizer
-import torch
-
-# Configuration du modèle
-device = "cuda" if torch.cuda.is_available() else "cpu"
-model = AutoModelForCausalLM.from_pretrained(
-    "soynade-research/Oolel-v0.1",
-    torch_dtype=torch.bfloat16,
-    device_map="auto" if torch.cuda.is_available() else None
-)
-tokenizer = AutoTokenizer.from_pretrained("soynade-research/Oolel-v0.1")
+import gradio as gr
 
-def generate_response(messages, max_new_tokens=1024, temperature=0.1):
-    text = tokenizer.apply_chat_template(
-        messages,
-        tokenize=False,
-        add_generation_prompt=True
-    )
-    model_inputs = tokenizer([text], return_tensors="pt").to(device)
-    generated_ids = model.generate(
-        model_inputs.input_ids, 
-        max_new_tokens=max_new_tokens, 
-        temperature=temperature
-    )
-    
-    generated_ids = [
-        output_ids[len(input_ids):] for input_ids, output_ids in zip(model_inputs.input_ids, generated_ids)
-    ]
-    response = tokenizer.batch_decode(generated_ids, skip_special_tokens=True)[0]
-    return response
+# Charger le modèle
+model_name = "soynade-research/Oolel-v0.1"
+tokenizer = AutoTokenizer.from_pretrained(model_name)
+model = AutoModelForCausalLM.from_pretrained(model_name, device_map="auto")
 
-# Configuration de l'interface Gradio
-def chat_interface(message, history):
-    # Convertir l'historique de Gradio au format requis par le modèle
-    formatted_history = [
-        {"role": "user" if idx % 2 == 0 else "assistant", "content": msg} 
-        for idx, msg in enumerate(sum(history, []))
-    ]
-    
-    # Ajouter le nouveau message
-    formatted_history.append({"role": "user", "content": message})
-    
-    # Générer la réponse
-    response = generate_response(formatted_history)
-    
-    return response
+# Fonction pour générer une réponse
+def generate_response(user_input, max_new_tokens=150, temperature=0.7):
+    inputs = tokenizer(user_input, return_tensors="pt").to("cuda")
+    outputs = model.generate(inputs.input_ids, max_new_tokens=max_new_tokens, temperature=temperature)
+    return tokenizer.decode(outputs[0], skip_special_tokens=True)
 
-# Créer l'interface Gradio
-iface = gr.ChatInterface(
-    fn=chat_interface,
-    title="Chat avec Oolel",
-    description="Conversez avec le modèle Oolel",
-    type="messages"
+# Interface Gradio
+iface = gr.Interface(
+    fn=generate_response,
+    inputs=[gr.Textbox(label="Message utilisateur"), gr.Slider(50, 500, value=150, label="Nombre max de tokens")],
+    outputs="text",
+    title="Oolel Chatbot"
 )
 
-if __name__ == "__main__":
-    iface.launch()
+iface.launch()