chunked/nltk_chunking/tasks_token_classification/chunk_13.txt

', '[SEP]']

However, this adds some special tokens [CLS] and [SEP] and the subword tokenization creates a mismatch between the input and labels.