[Bugfix] `OPTDecoderLayer` does not return attentions when...

[Bugfix] `OPTDecoderLayer` does not return attentions when `gradient_checkpointing` and `training` is enabled. (#23367)

Update modeling_opt.py

src/transformers/models/opt/modeling_opt.py

@@ -299,9 +299,9 @@ class OPTDecoderLayer(nn.Module):
         hidden_states: torch.Tensor,
         attention_mask: Optional[torch.Tensor] = None,
         layer_head_mask: Optional[torch.Tensor] = None,
+        past_key_value: Optional[Tuple[torch.Tensor]] = None,
         output_attentions: Optional[bool] = False,
         use_cache: Optional[bool] = False,
-        past_key_value: Optional[Tuple[torch.Tensor]] = None,
     ) -> Tuple[torch.FloatTensor, Optional[Tuple[torch.FloatTensor, torch.FloatTensor]]]:
         """
         Args: