Make create_extended_attention_mask_for_decoder static method (#16893)

src/transformers/modeling_utils.py

@@ -593,7 +593,8 @@ class ModuleUtilsMixin:
 
         return encoder_extended_attention_mask
 
-    def create_extended_attention_mask_for_decoder(self, input_shape, attention_mask, device):
+    @staticmethod
+    def create_extended_attention_mask_for_decoder(input_shape, attention_mask, device):
         batch_size, seq_length = input_shape
         seq_ids = torch.arange(seq_length, device=device)
         causal_mask = seq_ids[None, None, :].repeat(batch_size, seq_length, 1) <= seq_ids[None, :, None]
@@ -638,7 +639,7 @@ class ModuleUtilsMixin:
             # - if the model is a decoder, apply a causal mask in addition to the padding mask
             # - if the model is an encoder, make the mask broadcastable to [batch_size, num_heads, seq_length, seq_length]
             if self.config.is_decoder:
-                extended_attention_mask = self.create_extended_attention_mask_for_decoder(
+                extended_attention_mask = ModuleUtilsMixin.create_extended_attention_mask_for_decoder(
                     input_shape, attention_mask, device
                 )
             else: