from transformers import BertTokenizer # Load the BERT tokenizer. print('Loading BERT tokenizer...') tokenizer = BertTokenizer.from_pretrained('bert-base-uncased', do_lower_case=True)