accelerate tokenizer

tokenization_chatglm.py

@@ -439,5 +439,8 @@ class ChatGLMTokenizer(PreTrainedTokenizer):
                 encoded_inputs["position_ids"] = np.pad(encoded_inputs["position_ids"],
                                                         pad_width=[(0, 0), (difference, 0)])
             encoded_inputs[self.model_input_names[0]] = [self.pad_token_id] * difference + required_input
+        encoded_inputs["attention_mask"] = encoded_inputs["attention_mask"].tolist()
+        encoded_inputs["position_ids"] = encoded_inputs["position_ids"].tolist()
+
 
         return encoded_inputs