Use Python 3.9 syntax in tests (#37343)

Signed-off-by: cyy <cyyever@outlook.com>
This commit is contained in:
cyyever
2025-04-08 20:12:08 +08:00
committed by GitHub
parent 0fc683d1cd
commit 1e6b546ea6
666 changed files with 265 additions and 946 deletions

View File

@@ -1,4 +1,3 @@
# coding=utf-8
# Copyright 2021 The HuggingFace Inc. team.
#
# Licensed under the Apache License, Version 2.0 (the "License");
@@ -20,7 +19,6 @@ import shutil
import tempfile
import unittest
from functools import lru_cache
from typing import Tuple
from transformers import AddedToken, BatchEncoding, PerceiverTokenizer
from transformers.utils import cached_property, is_tf_available, is_torch_available
@@ -58,7 +56,7 @@ class PerceiverTokenizationTest(TokenizerTesterMixin, unittest.TestCase):
pretrained_name = pretrained_name or cls.tmpdirname
return cls.tokenizer_class.from_pretrained(pretrained_name, **kwargs)
def get_clean_sequence(self, tokenizer, with_prefix_space=False, max_length=20, min_length=5) -> Tuple[str, list]:
def get_clean_sequence(self, tokenizer, with_prefix_space=False, max_length=20, min_length=5) -> tuple[str, list]:
# XXX The default common tokenizer tests assume that every ID is decodable on its own.
# This assumption is invalid for Perceiver because single bytes might not be
# valid utf-8 (byte 128 for instance).