from transformers import AutoTokenizer

tokenizer = AutoTokenizer.from_pretrained("bert-base-uncased")
print(tokenizer._tokenizer.model)

# WordPiece(...)