vocab/dutch_llama_tokenizer/__init__.py

from transformers import AutoTokenizer

tokenizer = AutoTokenizer.from_pretrained("yhavinga/dutch-llama-tokenizer", trust_remote_code=True)