Update app.py

app.py

@@ -28,6 +28,11 @@ hf_utils = importlib.import_module('transformers.utils')
 hf_utils.is_flash_attn_2_available            = lambda *a, **k: False
 hf_utils.is_flash_attn_greater_or_equal_2_10 = lambda *a, **k: False
 
+mask_utils = importlib.import_module("transformers.modeling_attn_mask_utils")
+for fn in ("_prepare_4d_attention_mask_for_sdpa", "_prepare_4d_causal_attention_mask_for_sdpa"):
+    if not hasattr(mask_utils, fn):
+        setattr(mask_utils, fn, lambda *a, **k: None)
+
 REVISION = "ceaf371f01ef66192264811b390bccad475a4f02"
 
 # Florence-2 로드