Update app.py

app.py

@@ -13,11 +13,12 @@ import spaces
 processor = AutoProcessor.from_pretrained(
     "rhymes-ai/Aria",
     revision="0e2a8319",
+    torch_dtype=torch.bfloat16,
 )
 model = AutoModel.from_pretrained("rhymes-ai/Aria", 
         revision="f7c92bba",
         torch_dtype=torch.bfloat16,
-        #_attn_implementation="flash_attention_2"
+        attn_implementation="flash_attention_2"
         ).to("cuda")
 
 @spaces.GPU