update

modeling_flexbert.py

@@ -935,7 +935,7 @@ class FlexBertModel(FlexBertPreTrainedModel):
         else:
             self.final_norm = None
         self.unpad_embeddings = config.unpad_embeddings
-        self.is_decoder = config.get("is_casual", False)
+        self.is_decoder = False
 
     def post_init(self):
         self._init_weights(reset_params=False)
@@ -1515,6 +1515,7 @@ class FlexBertForCausalLM(FlexBertPreTrainedModel):
     def __init__(self, config: FlexBertConfig):
         super().__init__(config)
         self.bert = FlexBertModel(config)
+        self.bert.is_decoder = True
         self.lm_head = FlexBertPredictionHead(config)
 
         if config.tie_word_embeddings: