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

Fix typos (#36910)

Browse Source 
* fix typos

* fix typos

* fix typos

* fix typos
This commit is contained in:
omahs
2025-03-24 15:08:29 +01:00
committed by GitHub
parent 340500b1a9
commit cbf924b76c
67 changed files with 111 additions and 111 deletions
Download Patch File Download Diff File Expand all files Collapse all files

24
tests/models/llava/test_modeling_llava.py
View File

@@ -253,7 +253,7 @@ class LlavaForConditionalGenerationModelTest(ModelTesterMixin, GenerationTesterM
def test_mismatching_num_image_tokens(self):
"""
Tests that VLMs through an error with explicit message saying what is wrong
when number of images don't match number of image tokens in the text.
when number of images doesn't match number of image tokens in the text.
Also we need to test multi-image cases when one prompr has multiple image tokens.
"""
config, input_dict = self.model_tester.prepare_config_and_inputs_for_common()
@@ -306,19 +306,19 @@ class LlavaForConditionalGenerationModelTest(ModelTesterMixin, GenerationTesterM
model(**input_dict)
@unittest.skip(
reason="This architecture seem to not compute gradients properly when using GC, check: https://github.com/huggingface/transformers/pull/27124"
reason="This architecture seems to not compute gradients properly when using GC, check: https://github.com/huggingface/transformers/pull/27124"
)
def test_training_gradient_checkpointing(self):
pass
@unittest.skip(
reason="This architecture seem to not compute gradients properly when using GC, check: https://github.com/huggingface/transformers/pull/27124"
reason="This architecture seems to not compute gradients properly when using GC, check: https://github.com/huggingface/transformers/pull/27124"
)
def test_training_gradient_checkpointing_use_reentrant(self):
pass
@unittest.skip(
reason="This architecture seem to not compute gradients properly when using GC, check: https://github.com/huggingface/transformers/pull/27124"
reason="This architecture seems to not compute gradients properly when using GC, check: https://github.com/huggingface/transformers/pull/27124"
)
def test_training_gradient_checkpointing_use_reentrant_false(self):
pass
@@ -345,7 +345,7 @@ class LlavaForConditionalGenerationIntegrationTest(unittest.TestCase):
@slow
@require_bitsandbytes
def test_small_model_integration_test(self):
# Let' s make sure we test the preprocessing to replace what is used
# Let's make sure we test the preprocessing to replace what is used
model = LlavaForConditionalGeneration.from_pretrained("llava-hf/bakLlava-v1-hf", load_in_4bit=True)
prompt = "<image>\nUSER: What are the things I should be cautious about when I visit this place?\nASSISTANT:"
@@ -364,7 +364,7 @@ class LlavaForConditionalGenerationIntegrationTest(unittest.TestCase):
@slow
@require_bitsandbytes
def test_small_model_integration_test_llama_single(self):
# Let' s make sure we test the preprocessing to replace what is used
# Let's make sure we test the preprocessing to replace what is used
model_id = "llava-hf/llava-1.5-7b-hf"
model = LlavaForConditionalGeneration.from_pretrained("llava-hf/llava-1.5-7b-hf", load_in_4bit=True)
@@ -386,7 +386,7 @@ class LlavaForConditionalGenerationIntegrationTest(unittest.TestCase):
@slow
@require_bitsandbytes
def test_small_model_integration_test_llama_batched(self):
# Let' s make sure we test the preprocessing to replace what is used
# Let's make sure we test the preprocessing to replace what is used
model_id = "llava-hf/llava-1.5-7b-hf"
model = LlavaForConditionalGeneration.from_pretrained("llava-hf/llava-1.5-7b-hf", load_in_4bit=True)
@@ -413,7 +413,7 @@ class LlavaForConditionalGenerationIntegrationTest(unittest.TestCase):
@slow
@require_bitsandbytes
def test_small_model_integration_test_batch(self):
# Let' s make sure we test the preprocessing to replace what is used
# Let's make sure we test the preprocessing to replace what is used
model = LlavaForConditionalGeneration.from_pretrained("llava-hf/bakLlava-v1-hf", load_in_4bit=True)
# The first batch is longer in terms of text, but only has 1 image. The second batch will be padded in text, but the first will be padded because images take more space!.
prompts = [
@@ -441,7 +441,7 @@ class LlavaForConditionalGenerationIntegrationTest(unittest.TestCase):
@slow
@require_bitsandbytes
def test_small_model_integration_test_llama_batched_regression(self):
# Let' s make sure we test the preprocessing to replace what is used
# Let's make sure we test the preprocessing to replace what is used
model_id = "llava-hf/llava-1.5-7b-hf"
# Multi-image & multi-prompt (e.g. 3 images and 2 prompts now fails with SDPA, this tests if "eager" works as before)
@@ -478,7 +478,7 @@ class LlavaForConditionalGenerationIntegrationTest(unittest.TestCase):
processor = AutoProcessor.from_pretrained("llava-hf/llava-1.5-7b-hf")
prompt1 = "<image>\n<image>\nUSER: What's the the difference of two images?\nASSISTANT:"
prompt1 = "<image>\n<image>\nUSER: What's the difference of two images?\nASSISTANT:"
prompt2 = "<image>\nUSER: Describe the image.\nASSISTANT:"
prompt3 = "<image>\nUSER: Describe the image.\nASSISTANT:"
url1 = "https://images.unsplash.com/photo-1552053831-71594a27632d?q=80&w=3062&auto=format&fit=crop&ixlib=rb-4.0.3&ixid=M3wxMjA3fDB8MHxwaG90by1wYWdlfHx8fGVufDB8fHx8fA%3D%3D"
@@ -496,7 +496,7 @@ class LlavaForConditionalGenerationIntegrationTest(unittest.TestCase):
model = model.eval()
EXPECTED_OUTPUT = [
"\n \nUSER: What's the the difference of two images?\nASSISTANT: The difference between the two images is that one shows a dog standing on a grassy field, while",
"\n \nUSER: What's the difference of two images?\nASSISTANT: The difference between the two images is that one shows a dog standing on a grassy field, while",
"\nUSER: Describe the image.\nASSISTANT: The image features a brown and white dog sitting on a sidewalk. The dog is holding a small",
"\nUSER: Describe the image.\nASSISTANT: The image features a lone llama standing on a grassy hill. The llama is the",
]
@@ -617,7 +617,7 @@ These descriptions provide a detailed overview of the content and atmosphere of
generate_ids = model.generate(**inputs, max_new_tokens=50)
output = processor.batch_decode(generate_ids, skip_special_tokens=True, clean_up_tokenization_spaces=False)[0]
EXPECTED_GENERATION = "Describe the images.The image showcases a dog, which is prominently positioned in the center, taking up a significant portion of the frame. The dog is situated against a backdrop of a wooden surface, which spans the entire image. The dog appears to be a black Labrador" # fmt: skip
EXPECTED_GENERATION = "Describe the images. The image showcases a dog, which is prominently positioned in the center, taking up a significant portion of the frame. The dog is situated against a backdrop of a wooden surface, which spans the entire image. The dog appears to be a black Labrador" # fmt: skip
self.assertEqual(output, EXPECTED_GENERATION)
@slow