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

BLIPs clean-up (#35560)

Browse Source 
* blips clean up

* update processor

* readability

* fix processor length

* fix copies

* tmp

* update and fix copies

* why keep these, delete?

* fix test fetcher

* irrelevant comment

* fix tests

* fix tests

* fix copies
This commit is contained in:
Raushan Turganbay
2025-07-29 10:03:06 +02:00
committed by GitHub
parent 4f8f51be4e
commit 75794792ad
15 changed files with 301 additions and 592 deletions
Download Patch File Download Diff File Expand all files Collapse all files

39
tests/models/instructblipvideo/test_processor_instructblipvideo.py
View File

@@ -59,6 +59,9 @@ class InstructBlipVideoProcessorTest(ProcessorTesterMixin, unittest.TestCase):
def get_qformer_tokenizer(self, **kwargs):
return AutoProcessor.from_pretrained(self.tmpdirname, **kwargs).qformer_tokenizer
def prepare_processor_dict(self):
return {"num_query_tokens": 1}
def get_video_processor(self, **kwargs):
return AutoProcessor.from_pretrained(self.tmpdirname, **kwargs).video_processor
@@ -93,9 +96,13 @@ class InstructBlipVideoProcessorTest(ProcessorTesterMixin, unittest.TestCase):
video_processor = self.get_video_processor()
tokenizer = self.get_tokenizer()
qformer_tokenizer = self.get_qformer_tokenizer()
processor_kwargs = self.prepare_processor_dict()
processor = InstructBlipVideoProcessor(
tokenizer=tokenizer, video_processor=video_processor, qformer_tokenizer=qformer_tokenizer
tokenizer=tokenizer,
video_processor=video_processor,
qformer_tokenizer=qformer_tokenizer,
**processor_kwargs,
)
image_input = self.prepare_image_inputs()
@@ -110,15 +117,17 @@ class InstructBlipVideoProcessorTest(ProcessorTesterMixin, unittest.TestCase):
video_processor = self.get_video_processor()
tokenizer = self.get_tokenizer()
qformer_tokenizer = self.get_qformer_tokenizer()
processor_kwargs = self.prepare_processor_dict()
processor = InstructBlipVideoProcessor(
tokenizer=tokenizer, video_processor=video_processor, qformer_tokenizer=qformer_tokenizer
tokenizer=tokenizer,
video_processor=video_processor,
qformer_tokenizer=qformer_tokenizer,
**processor_kwargs,
)
input_str = ["lower newer"]
encoded_processor = processor(text=input_str)
encoded_tokens = tokenizer(input_str, return_token_type_ids=False)
encoded_tokens_qformer = qformer_tokenizer(input_str, return_token_type_ids=False)
@@ -132,9 +141,13 @@ class InstructBlipVideoProcessorTest(ProcessorTesterMixin, unittest.TestCase):
video_processor = self.get_video_processor()
tokenizer = self.get_tokenizer()
qformer_tokenizer = self.get_qformer_tokenizer()
processor_kwargs = self.prepare_processor_dict()
processor = InstructBlipVideoProcessor(
tokenizer=tokenizer, video_processor=video_processor, qformer_tokenizer=qformer_tokenizer
tokenizer=tokenizer,
video_processor=video_processor,
qformer_tokenizer=qformer_tokenizer,
**processor_kwargs,
)
input_str = "lower newer"
@@ -144,7 +157,7 @@ class InstructBlipVideoProcessorTest(ProcessorTesterMixin, unittest.TestCase):
self.assertListEqual(
list(inputs.keys()),
["input_ids", "attention_mask", "qformer_input_ids", "qformer_attention_mask", "pixel_values"],
["qformer_input_ids", "qformer_attention_mask", "input_ids", "attention_mask", "pixel_values"],
)
# test if it raises when no input is passed
@@ -155,9 +168,13 @@ class InstructBlipVideoProcessorTest(ProcessorTesterMixin, unittest.TestCase):
video_processor = self.get_video_processor()
tokenizer = self.get_tokenizer()
qformer_tokenizer = self.get_qformer_tokenizer()
processor_kwargs = self.prepare_processor_dict()
processor = InstructBlipVideoProcessor(
tokenizer=tokenizer, video_processor=video_processor, qformer_tokenizer=qformer_tokenizer
tokenizer=tokenizer,
video_processor=video_processor,
qformer_tokenizer=qformer_tokenizer,
**processor_kwargs,
)
predicted_ids = [[1, 4, 5, 8, 1, 0, 8], [3, 4, 3, 1, 1, 8, 9]]
@@ -171,9 +188,13 @@ class InstructBlipVideoProcessorTest(ProcessorTesterMixin, unittest.TestCase):
video_processor = self.get_video_processor()
tokenizer = self.get_tokenizer()
qformer_tokenizer = self.get_qformer_tokenizer()
processor_kwargs = self.prepare_processor_dict()
processor = InstructBlipVideoProcessor(
tokenizer=tokenizer, video_processor=video_processor, qformer_tokenizer=qformer_tokenizer
tokenizer=tokenizer,
video_processor=video_processor,
qformer_tokenizer=qformer_tokenizer,
**processor_kwargs,
)
input_str = "lower newer"
@@ -183,5 +204,5 @@ class InstructBlipVideoProcessorTest(ProcessorTesterMixin, unittest.TestCase):
self.assertListEqual(
list(inputs.keys()),
["input_ids", "attention_mask", "qformer_input_ids", "qformer_attention_mask", "pixel_values"],
["qformer_input_ids", "qformer_attention_mask", "input_ids", "attention_mask", "pixel_values"],
)