ahans30's picture
initial commit
d389c0e
raw
history blame contribute delete
No virus
409 Bytes
from transformers import AutoTokenizer
def assert_tokenizer_consistency(model_id_1, model_id_2):
identical_tokenizers = (
AutoTokenizer.from_pretrained(model_id_1).vocab
== AutoTokenizer.from_pretrained(model_id_2).vocab
)
if not identical_tokenizers:
raise ValueError(f"Tokenizers are not identical for {model_id_1.name_of_path} and {model_id_2.name_of_path}.")