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

🚨 🚨 Setup -> setupclass conversion (#37282)

Browse Source 
* More limited setup -> setupclass conversion

* make fixup

* Trigger tests

* Fixup UDOP

* Missed a spot

* tearDown -> tearDownClass where appropriate

* Couple more class fixes

* Fixups for UDOP and VisionTextDualEncoder

* Ignore errors when removing the tmpdir, in case it already got cleaned up somewhere

* CLIP fixes

* More correct classmethods

* Wav2Vec2Bert fixes

* More methods become static

* More class methods

* More class methods

* Revert changes for integration tests / modeling files

* Use a different tempdir for tests that actually write to it

* Remove addClassCleanup and just use teardownclass

* Remove changes in modeling files

* Cleanup get_processor_dict() for got_ocr2

* Fix regression on Wav2Vec2BERT test that was masked by this before

* Rework tests that modify the tmpdir

* make fix-copies

* revert clvp modeling test changes

* Fix CLIP processor test

* make fix-copies
This commit is contained in:
Matt
2025-04-08 17:15:37 +01:00
committed by GitHub
parent c15a7adb28
commit 4d0de5f73a
49 changed files with 740 additions and 574 deletions
Download Patch File Download Diff File Expand all files Collapse all files

78
tests/models/grounding_dino/test_processor_grounding_dino.py
View File

@@ -44,12 +44,13 @@ class GroundingDinoProcessorTest(ProcessorTesterMixin, unittest.TestCase):
from_pretrained_id = "IDEA-Research/grounding-dino-base"
processor_class = GroundingDinoProcessor
def setUp(self):
self.tmpdirname = tempfile.mkdtemp()
@classmethod
def setUpClass(cls):
cls.tmpdirname = tempfile.mkdtemp()
vocab_tokens = ["[UNK]","[CLS]","[SEP]","[PAD]","[MASK]","want","##want","##ed","wa","un","runn","##ing",",","low","lowest"] # fmt: skip
self.vocab_file = os.path.join(self.tmpdirname, VOCAB_FILES_NAMES["vocab_file"])
with open(self.vocab_file, "w", encoding="utf-8") as vocab_writer:
cls.vocab_file = os.path.join(cls.tmpdirname, VOCAB_FILES_NAMES["vocab_file"])
with open(cls.vocab_file, "w", encoding="utf-8") as vocab_writer:
vocab_writer.write("".join([x + "\n" for x in vocab_tokens]))
image_processor_map = {
@@ -62,21 +63,21 @@ class GroundingDinoProcessorTest(ProcessorTesterMixin, unittest.TestCase):
"rescale_factor": 1 / 255,
"do_pad": True,
}
self.image_processor_file = os.path.join(self.tmpdirname, IMAGE_PROCESSOR_NAME)
with open(self.image_processor_file, "w", encoding="utf-8") as fp:
cls.image_processor_file = os.path.join(cls.tmpdirname, IMAGE_PROCESSOR_NAME)
with open(cls.image_processor_file, "w", encoding="utf-8") as fp:
json.dump(image_processor_map, fp)
image_processor = GroundingDinoImageProcessor()
tokenizer = BertTokenizer.from_pretrained(self.from_pretrained_id)
tokenizer = BertTokenizer.from_pretrained(cls.from_pretrained_id)
processor = GroundingDinoProcessor(image_processor, tokenizer)
processor.save_pretrained(self.tmpdirname)
processor.save_pretrained(cls.tmpdirname)
self.batch_size = 7
self.num_queries = 5
self.embed_dim = 5
self.seq_length = 5
cls.batch_size = 7
cls.num_queries = 5
cls.embed_dim = 5
cls.seq_length = 5
def prepare_text_inputs(self, batch_size: Optional[int] = None):
labels = ["a cat", "remote control"]
@@ -92,21 +93,24 @@ class GroundingDinoProcessorTest(ProcessorTesterMixin, unittest.TestCase):
return [labels]
return [labels, labels_longer] + [labels] * (batch_size - 2)
@classmethod
# Copied from tests.models.clip.test_processor_clip.CLIPProcessorTest.get_tokenizer with CLIP->Bert
def get_tokenizer(self, **kwargs):
return BertTokenizer.from_pretrained(self.tmpdirname, **kwargs)
def get_tokenizer(cls, **kwargs):
return BertTokenizer.from_pretrained(cls.tmpdirname, **kwargs)
@classmethod
# Copied from tests.models.clip.test_processor_clip.CLIPProcessorTest.get_rust_tokenizer with CLIP->Bert
def get_rust_tokenizer(self, **kwargs):
return BertTokenizerFast.from_pretrained(self.tmpdirname, **kwargs)
def get_rust_tokenizer(cls, **kwargs):
return BertTokenizerFast.from_pretrained(cls.tmpdirname, **kwargs)
@classmethod
# Copied from tests.models.clip.test_processor_clip.CLIPProcessorTest.get_image_processor with CLIP->GroundingDino
def get_image_processor(self, **kwargs):
return GroundingDinoImageProcessor.from_pretrained(self.tmpdirname, **kwargs)
def get_image_processor(cls, **kwargs):
return GroundingDinoImageProcessor.from_pretrained(cls.tmpdirname, **kwargs)
# Copied from tests.models.clip.test_processor_clip.CLIPProcessorTest.tearDown
def tearDown(self):
shutil.rmtree(self.tmpdirname)
@classmethod
def tearDownClass(cls):
shutil.rmtree(cls.tmpdirname, ignore_errors=True)
def get_fake_grounding_dino_output(self):
torch.manual_seed(42)
@@ -147,13 +151,14 @@ class GroundingDinoProcessorTest(ProcessorTesterMixin, unittest.TestCase):
tokenizer_fast = self.get_rust_tokenizer()
image_processor = self.get_image_processor()
processor_slow = GroundingDinoProcessor(tokenizer=tokenizer_slow, image_processor=image_processor)
processor_slow.save_pretrained(self.tmpdirname)
processor_slow = GroundingDinoProcessor.from_pretrained(self.tmpdirname, use_fast=False)
with tempfile.TemporaryDirectory() as tmpdir:
processor_slow = GroundingDinoProcessor(tokenizer=tokenizer_slow, image_processor=image_processor)
processor_slow.save_pretrained(tmpdir)
processor_slow = GroundingDinoProcessor.from_pretrained(tmpdir, use_fast=False)
processor_fast = GroundingDinoProcessor(tokenizer=tokenizer_fast, image_processor=image_processor)
processor_fast.save_pretrained(self.tmpdirname)
processor_fast = GroundingDinoProcessor.from_pretrained(self.tmpdirname)
processor_fast = GroundingDinoProcessor(tokenizer=tokenizer_fast, image_processor=image_processor)
processor_fast.save_pretrained(tmpdir)
processor_fast = GroundingDinoProcessor.from_pretrained(tmpdir)
self.assertEqual(processor_slow.tokenizer.get_vocab(), tokenizer_slow.get_vocab())
self.assertEqual(processor_fast.tokenizer.get_vocab(), tokenizer_fast.get_vocab())
@@ -168,15 +173,20 @@ class GroundingDinoProcessorTest(ProcessorTesterMixin, unittest.TestCase):
# Copied from tests.models.clip.test_processor_clip.CLIPProcessorTest.test_save_load_pretrained_additional_features with CLIP->GroundingDino,GroundingDinoTokenizer->BertTokenizer
def test_save_load_pretrained_additional_features(self):
processor = GroundingDinoProcessor(tokenizer=self.get_tokenizer(), image_processor=self.get_image_processor())
processor.save_pretrained(self.tmpdirname)
with tempfile.TemporaryDirectory() as tmpdir:
processor = GroundingDinoProcessor(
tokenizer=self.get_tokenizer(), image_processor=self.get_image_processor()
)
processor.save_pretrained(tmpdir)
tokenizer_add_kwargs = self.get_tokenizer(bos_token="(BOS)", eos_token="(EOS)")
image_processor_add_kwargs = self.get_image_processor(do_normalize=False, padding_value=1.0)
tokenizer_add_kwargs = BertTokenizer.from_pretrained(tmpdir, bos_token="(BOS)", eos_token="(EOS)")
image_processor_add_kwargs = GroundingDinoImageProcessor.from_pretrained(
tmpdir, do_normalize=False, padding_value=1.0
)
processor = GroundingDinoProcessor.from_pretrained(
self.tmpdirname, bos_token="(BOS)", eos_token="(EOS)", do_normalize=False, padding_value=1.0
)
processor = GroundingDinoProcessor.from_pretrained(
tmpdir, bos_token="(BOS)", eos_token="(EOS)", do_normalize=False, padding_value=1.0
)
self.assertEqual(processor.tokenizer.get_vocab(), tokenizer_add_kwargs.get_vocab())
self.assertIsInstance(processor.tokenizer, BertTokenizerFast)