How to join sub words produced by the named entity recognization task on transformer huggingface?

Question

from transformers import pipeline, AutoModelForTokenClassification, AutoTokenizer

model = AutoModelForTokenClassification.from_pretrained("/nfs/storages/bio_corpus