SUMIN/HuggingFace_transformer
1
0
Fork 0
Code Issues Pull Requests Actions 7 Packages Projects Releases Wiki Activity

SPLIT PR: add user defined symbols and control symbols (#31305)

Browse Source 
* PR SPLIT: moving origina changes for adding user defined symbols

* adding gemma test and generalizing gemma converter

* ruff

* update common test

* update serialization test

* deberta v2 tests updates as rust version adds '.' as a user added token, so a space is not added

* removing commented lines

* applying feedback - user only added_tokens to add and check piece.type instead of trainer_spec for user_defined_symbols

* add comment referencing sentencepiece
This commit is contained in:
Ita Zaporozhets
2024-06-21 10:48:10 +02:00
committed by GitHub
parent 730a440734
commit 1e79eade41
6 changed files with 60 additions and 23 deletions
Download Patch File Download Diff File Expand all files Collapse all files

9
tests/models/rembert/test_tokenization_rembert.py
View File

@@ -172,7 +172,7 @@ class RemBertTokenizationTest(TokenizerTesterMixin, unittest.TestCase):
self.assertTrue(str(expected_eos) not in tokenizer.additional_special_tokens)
self.assertIn(new_eos, tokenizer.added_tokens_decoder.values())
self.assertEqual(tokenizer.added_tokens_decoder[tokenizer.eos_token_id], new_eos)
self.assertDictEqual(expected, tokenizer.added_tokens_decoder)
self.assertTrue(all(item in tokenizer.added_tokens_decoder.items() for item in expected.items()))
return tokenizer
new_eos = AddedToken("[NEW_EOS]", rstrip=False, lstrip=True, normalized=False, special=True)
@@ -227,7 +227,12 @@ class RemBertTokenizationTest(TokenizerTesterMixin, unittest.TestCase):
self.assertIn(new_eos, list(tokenizer_fast.added_tokens_decoder.values()))
# We can't test the following because for BC we kept the default rstrip lstrip in slow not fast. Will comment once normalization is alright
with self.subTest("Hub -> Fast == Hub -> Slow: make sure slow and fast tokenizer match"):
self.assertDictEqual(EXPECTED_ADDED_TOKENS_DECODER, tokenizer_fast.added_tokens_decoder)
self.assertTrue(
all(
item in tokenizer.added_tokens_decoder.items()
for item in EXPECTED_ADDED_TOKENS_DECODER.items()
)
)
EXPECTED_ADDED_TOKENS_DECODER = tokenizer_fast.added_tokens_decoder
with tempfile.TemporaryDirectory() as tmp_dir_4: