Get started by loading a pretrained tokenizer with the [AutoTokenizer.from_pretrained] method.