A problem about FutureWarning?

when a user my vocab to pretrain, my code is

tokenizer = BertTokenizer.from_pretrained('my_vocab.txt' )

the out is:
FutureWarning: This dataset will be removed from the library soon, preprocessing should be handled with the :hugs: Datasets library. You can have a look at this example script for pointers: transformers/run_mlm.py at master · huggingface/transformers · GitHub

how can i do is corect in future? I don`t understand this.