Update modeling_caption_bert.py

modeling_caption_bert.py

@@ -82,7 +82,8 @@ class CaptionBertModel(PreTrainedModel):
             norm_first=True,
         )
         self.encoder = nn.TransformerEncoder(
-            encoder_layer, num_layers=config.num_hidden_layers)
+            encoder_layer, num_layers=config.num_hidden_layers,
+            enable_nested_tensor=False)
 
         # Output projection to consensus space
         self.output_proj = nn.Sequential(