Use lru_cache for tokenization tests (#36818)

* fix

* fix

* fix

* fix

---------

Co-authored-by: ydshieh <ydshieh@users.noreply.github.com>

tests/models/flaubert/test_tokenization_flaubert.py

@@ -30,8 +30,9 @@ class FlaubertTokenizationTest(TokenizerTesterMixin, unittest.TestCase):
     tokenizer_class = FlaubertTokenizer
     test_rust_tokenizer = False
 
-    def setUp(self):
-        super().setUp()
+    @classmethod
+    def setUpClass(cls):
+        super().setUpClass()
 
         # Adapted from Sennrich et al. 2015 and https://github.com/rsennrich/subword-nmt
         vocab = ["l", "o", "w", "e", "r", "s", "t", "i", "d", "n", "w</w>", "r</w>", "t</w>", "i</w>", "lo", "low", "ne", "new", "er</w>", "low</w>", "lowest</w>", "new</w>", "newer</w>", "wider</w>", "<unk>"]  # fmt: skip
@@ -39,11 +40,11 @@ class FlaubertTokenizationTest(TokenizerTesterMixin, unittest.TestCase):
         vocab_tokens = dict(zip(vocab, range(len(vocab))))
         merges = ["n e 300", "ne w 301", "e r</w> 302", ""]
 
-        self.vocab_file = os.path.join(self.tmpdirname, VOCAB_FILES_NAMES["vocab_file"])
-        self.merges_file = os.path.join(self.tmpdirname, VOCAB_FILES_NAMES["merges_file"])
-        with open(self.vocab_file, "w", encoding="utf-8") as fp:
+        cls.vocab_file = os.path.join(cls.tmpdirname, VOCAB_FILES_NAMES["vocab_file"])
+        cls.merges_file = os.path.join(cls.tmpdirname, VOCAB_FILES_NAMES["merges_file"])
+        with open(cls.vocab_file, "w", encoding="utf-8") as fp:
             fp.write(json.dumps(vocab_tokens) + "\n")
-        with open(self.merges_file, "w", encoding="utf-8") as fp:
+        with open(cls.merges_file, "w", encoding="utf-8") as fp:
             fp.write("\n".join(merges))
 
     # Copied from transformers.tests.models.xlm.test_tokenization_xlm.XLMTokenizationTest.test_full_tokenizer