Update modeling_chatglm.py

modeling_chatglm.py

@@ -860,6 +860,9 @@ class ChatGLMModel(ChatGLMPreTrainedModel):
 
         all_self_attentions = None
         all_hidden_states = () if output_hidden_states else None
+
+        hidden_states = inputs_embeds
+        
         for index, layer in enumerate(self.layers):
             if output_hidden_states:
                 all_hidden_states = all_hidden_states + (hidden_states,)