diff --git a/docs/data_prepare/data_preprocess.rst b/docs/data_prepare/data_preprocess.rst
index 338dae36cf17..cebf4102e2ec 100644
--- a/docs/data_prepare/data_preprocess.rst
+++ b/docs/data_prepare/data_preprocess.rst
@@ -75,7 +75,7 @@ paddlenlp内置的 :class:`paddlenlp.datasets.MapDataset` 的 :func:`map` 方法
     def convert_examples(examples, tokenizer):
         querys = [example['query'] for example in examples]
         titles = [example['title'] for example in examples]
-        tokenized_examples = tokenizer(text=querys, text_pair=titles)
+        tokenized_examples = tokenizer(text=querys, text_pair=titles,return_dict=False)
 
         # 加上label用于训练
         for idx in range(len(tokenized_examples)):