update tokenizer for compatibility with new transformers

tokenization_chatglm.py

@@ -271,6 +271,8 @@ class ChatGLMTokenizer(PreTrainedTokenizer):
         padding_strategy: PaddingStrategy = PaddingStrategy.DO_NOT_PAD,
         pad_to_multiple_of: Optional[int] = None,
         return_attention_mask: Optional[bool] = None,
+        padding_side: Optional[bool] = None,
+        **kwargs
     ) -> dict:
         """
         Pad encoded inputs (on left/right and up to predefined length or max length in the batch)
@@ -297,6 +299,7 @@ class ChatGLMTokenizer(PreTrainedTokenizer):
         """
         # Load from model defaults
         assert self.padding_side == "left"
+        assert padding_side is None or padding_side == "left"
 
         required_input = encoded_inputs[self.model_input_names[0]]
         seq_length = len(required_input)