add soynade-research/Oolel-v0.1

app.py

@@ -1,7 +1,58 @@
+# Fichier app.py
 import gradio as gr
+from transformers import AutoModelForCausalLM, AutoTokenizer
+import torch
 
-def greet(name):
-    return "Hello " + name + "!!"
+# 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")
 
-demo = gr.Interface(fn=greet, inputs="text", outputs="text")
-demo.launch()
+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
+
+# 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
+
+# Créer l'interface Gradio
+iface = gr.ChatInterface(
+    fn=chat_interface,
+    title="Chat avec Oolel",
+    description="Conversez avec le modèle Oolel",
+)
+
+if __name__ == "__main__":
+    iface.launch()

requirements.txt

@@ -0,0 +1,4 @@
+torch
+transformers
+gradio
+accelerate