Fix for transformers update

modeling_florence2.py

@@ -2340,7 +2340,9 @@ class Florence2PreTrainedModel(PreTrainedModel):
         Retrieve language_model's attribute to check whether the model supports
         Flash Attention 2 or not.
         """
-        return self.language_model._supports_flash_attn_2
+        if hasattr(self, 'language_model') and self.language_model is not None:
+            return self.language_model._supports_flash_attn_2
+        return True  # Default to True during initialization
 
     @property
     def _supports_sdpa(self):
@@ -2348,7 +2350,9 @@ class Florence2PreTrainedModel(PreTrainedModel):
         Retrieve language_model's attribute to check whether the model supports
         SDPA or not.
         """
-        return self.language_model._supports_sdpa
+        if hasattr(self, 'language_model') and self.language_model is not None:
+            return self.language_model._supports_sdpa
+        return True  # Default to True during initialization
 
 
 FLORENCE2_INPUTS_DOCSTRING = r"""

nodes.py

@@ -384,6 +384,7 @@ class Florence2Run:
                 max_new_tokens=max_new_tokens,
                 do_sample=do_sample,
                 num_beams=num_beams,
+                use_cache=False
             )
 
             results = processor.batch_decode(generated_ids, skip_special_tokens=False)[0]