Update app.py

app.py

@@ -1,30 +1,36 @@
 from fastapi import FastAPI, Request
 from fastapi.responses import StreamingResponse, HTMLResponse
-from fastapi.staticfiles import StaticFiles
 from fastapi.templating import Jinja2Templates
 from transformers import AutoTokenizer, AutoModelForCausalLM
 import torch
 import asyncio
+import os
+
+# Créer le dossier static s'il n'existe pas
+os.makedirs("static", exist_ok=True)
 
 app = FastAPI()
-app.mount("/static", StaticFiles(directory="static"), name="static")
 templates = Jinja2Templates(directory="templates")
 
-# Charger le modèle public (sans token)
+# Configuration simplifiée pour Hugging Face Spaces
 model_name = "microsoft/Phi-3.5-mini-instruct"
-tokenizer = AutoTokenizer.from_pretrained(model_name)
-model = AutoModelForCausalLM.from_pretrained(
-    model_name,
-    torch_dtype=torch.float16,
-    device_map="auto"
-)
+
+try:
+    tokenizer = AutoTokenizer.from_pretrained(model_name)
+    model = AutoModelForCausalLM.from_pretrained(
+        model_name,
+        torch_dtype=torch.float16 if torch.cuda.is_available() else torch.float32,
+        device_map="auto"
+    )
+except Exception as e:
+    print(f"Erreur de chargement du modèle: {str(e)}")
+    raise
 
 async def generate_response(prompt: str):
-    inputs = tokenizer(prompt, return_tensors="pt").to(model.device)
-    
-    # Génération token par token
-    with torch.no_grad():
-        for _ in range(512):  # Limite de tokens
+    try:
+        inputs = tokenizer(prompt, return_tensors="pt").to(model.device)
+        
+        for _ in range(512):
             outputs = model.generate(
                 **inputs,
                 max_new_tokens=1,
@@ -35,8 +41,9 @@ async def generate_response(prompt: str):
             new_token = tokenizer.decode(outputs[0][-1], skip_special_tokens=True)
             yield f"data: {new_token}\n\n"
             await asyncio.sleep(0.05)
-            
             inputs = {"input_ids": outputs}
+    except Exception as e:
+        yield f"data: [ERREUR: {str(e)}]\n\n"
 
 @app.get("/", response_class=HTMLResponse)
 async def home(request: Request):