Replace video_fps with fps in tests (#39898)
Signed-off-by: cyy <cyyever@outlook.com>
This commit is contained in:
@@ -234,7 +234,7 @@ class InternVLProcessorTest(ProcessorTesterMixin, unittest.TestCase):
|
||||
self.assertTrue(self.videos_input_name in out_dict_with_video)
|
||||
self.assertEqual(len(out_dict_with_video[self.videos_input_name]), num_frames)
|
||||
|
||||
# Load with `video_fps` arg is not possible with InternVL (skip)
|
||||
# Load with `fps` arg is not possible with InternVL (skip)
|
||||
|
||||
# Load without any arg should use the default loading method
|
||||
out_dict_with_video = processor.apply_chat_template(
|
||||
|
||||
@@ -488,26 +488,26 @@ class Qwen2_5OmniProcessorTest(ProcessorTesterMixin, unittest.TestCase):
|
||||
self.assertTrue(self.videos_input_name in out_dict_with_video)
|
||||
self.assertEqual(len(out_dict_with_video[self.videos_input_name]), 5760)
|
||||
|
||||
# Load with `video_fps` arg
|
||||
video_fps = 1
|
||||
# Load with `fps` arg
|
||||
fps = 1
|
||||
out_dict_with_video = processor.apply_chat_template(
|
||||
messages,
|
||||
add_generation_prompt=True,
|
||||
tokenize=True,
|
||||
return_dict=True,
|
||||
video_fps=video_fps,
|
||||
fps=fps,
|
||||
)
|
||||
self.assertTrue(self.videos_input_name in out_dict_with_video)
|
||||
self.assertEqual(len(out_dict_with_video[self.videos_input_name]), 14400)
|
||||
|
||||
# Load with `video_fps` and `num_frames` args, should raise an error
|
||||
# Load with `fps` and `num_frames` args, should raise an error
|
||||
with self.assertRaises(ValueError):
|
||||
out_dict_with_video = processor.apply_chat_template(
|
||||
messages,
|
||||
add_generation_prompt=True,
|
||||
tokenize=True,
|
||||
return_dict=True,
|
||||
video_fps=video_fps,
|
||||
fps=fps,
|
||||
num_frames=num_frames,
|
||||
)
|
||||
|
||||
|
||||
@@ -304,26 +304,26 @@ class Qwen2_5_VLProcessorTest(ProcessorTesterMixin, unittest.TestCase):
|
||||
self.assertTrue(self.videos_input_name in out_dict_with_video)
|
||||
self.assertEqual(len(out_dict_with_video[self.videos_input_name]), 360)
|
||||
|
||||
# Load with `video_fps` arg
|
||||
video_fps = 1
|
||||
# Load with `fps` arg
|
||||
fps = 1
|
||||
out_dict_with_video = processor.apply_chat_template(
|
||||
messages,
|
||||
add_generation_prompt=True,
|
||||
tokenize=True,
|
||||
return_dict=True,
|
||||
video_fps=video_fps,
|
||||
fps=fps,
|
||||
)
|
||||
self.assertTrue(self.videos_input_name in out_dict_with_video)
|
||||
self.assertEqual(len(out_dict_with_video[self.videos_input_name]), 900)
|
||||
|
||||
# Load with `video_fps` and `num_frames` args, should raise an error
|
||||
# Load with `fps` and `num_frames` args, should raise an error
|
||||
with self.assertRaises(ValueError):
|
||||
out_dict_with_video = processor.apply_chat_template(
|
||||
messages,
|
||||
add_generation_prompt=True,
|
||||
tokenize=True,
|
||||
return_dict=True,
|
||||
video_fps=video_fps,
|
||||
fps=fps,
|
||||
num_frames=num_frames,
|
||||
)
|
||||
|
||||
|
||||
@@ -304,26 +304,26 @@ class Qwen2VLProcessorTest(ProcessorTesterMixin, unittest.TestCase):
|
||||
self.assertTrue(self.videos_input_name in out_dict_with_video)
|
||||
self.assertEqual(len(out_dict_with_video[self.videos_input_name]), 360)
|
||||
|
||||
# Load with `video_fps` arg
|
||||
video_fps = 1
|
||||
# Load with `fps` arg
|
||||
fps = 1
|
||||
out_dict_with_video = processor.apply_chat_template(
|
||||
messages,
|
||||
add_generation_prompt=True,
|
||||
tokenize=True,
|
||||
return_dict=True,
|
||||
video_fps=video_fps,
|
||||
fps=fps,
|
||||
)
|
||||
self.assertTrue(self.videos_input_name in out_dict_with_video)
|
||||
self.assertEqual(len(out_dict_with_video[self.videos_input_name]), 900)
|
||||
|
||||
# Load with `video_fps` and `num_frames` args, should raise an error
|
||||
# Load with `fps` and `num_frames` args, should raise an error
|
||||
with self.assertRaises(ValueError):
|
||||
out_dict_with_video = processor.apply_chat_template(
|
||||
messages,
|
||||
add_generation_prompt=True,
|
||||
tokenize=True,
|
||||
return_dict=True,
|
||||
video_fps=video_fps,
|
||||
fps=fps,
|
||||
num_frames=num_frames,
|
||||
)
|
||||
|
||||
|
||||
@@ -417,20 +417,20 @@ class SmolVLMProcessorTest(ProcessorTesterMixin, unittest.TestCase):
|
||||
# SmolVLM doesn't sample `num_frames` exactly, by uses other sampling method
|
||||
self.assertEqual(len(out_dict_with_video[self.videos_input_name][0]), 3)
|
||||
|
||||
# Load with `video_fps` arg
|
||||
video_fps = 1
|
||||
# Load with `fps` arg
|
||||
fps = 1
|
||||
out_dict_with_video = processor.apply_chat_template(
|
||||
messages,
|
||||
add_generation_prompt=True,
|
||||
tokenize=True,
|
||||
return_dict=True,
|
||||
video_fps=video_fps,
|
||||
fps=fps,
|
||||
return_tensors="pt",
|
||||
)
|
||||
self.assertTrue(self.videos_input_name in out_dict_with_video)
|
||||
self.assertEqual(len(out_dict_with_video[self.videos_input_name]), 1)
|
||||
# SmolVLM doesn't sample 1 frame per second exactly, by uses other sampling method
|
||||
self.assertEqual(len(out_dict_with_video[self.videos_input_name][0]), video_fps * 10)
|
||||
self.assertEqual(len(out_dict_with_video[self.videos_input_name][0]), fps * 10)
|
||||
|
||||
# NOTE: the last assert checks are removed
|
||||
# Loading video as a list of frames (i.e. images) is not supported in SmolVLM
|
||||
|
||||
@@ -991,43 +991,43 @@ class ProcessorTesterMixin:
|
||||
self.assertEqual(len(out_dict_with_video[self.videos_input_name]), 1)
|
||||
self.assertEqual(len(out_dict_with_video[self.videos_input_name][0]), num_frames)
|
||||
|
||||
# Load with `video_fps` arg
|
||||
video_fps = 1
|
||||
# Load with `fps` arg
|
||||
fps = 1
|
||||
out_dict_with_video = processor.apply_chat_template(
|
||||
messages,
|
||||
add_generation_prompt=True,
|
||||
tokenize=True,
|
||||
return_dict=True,
|
||||
video_fps=video_fps,
|
||||
fps=fps,
|
||||
return_tensors="pt",
|
||||
)
|
||||
self.assertTrue(self.videos_input_name in out_dict_with_video)
|
||||
self.assertEqual(len(out_dict_with_video[self.videos_input_name]), 1)
|
||||
self.assertEqual(len(out_dict_with_video[self.videos_input_name][0]), video_fps * 10)
|
||||
self.assertEqual(len(out_dict_with_video[self.videos_input_name][0]), fps * 10)
|
||||
|
||||
# Whan `do_sample_frames=False` no sampling is done and whole video is loaded, even if number of frames is passed
|
||||
video_fps = 1
|
||||
fps = 1
|
||||
out_dict_with_video = processor.apply_chat_template(
|
||||
messages,
|
||||
add_generation_prompt=True,
|
||||
tokenize=True,
|
||||
return_dict=True,
|
||||
do_sample_frames=False,
|
||||
video_fps=video_fps,
|
||||
fps=fps,
|
||||
return_tensors="pt",
|
||||
)
|
||||
self.assertTrue(self.videos_input_name in out_dict_with_video)
|
||||
self.assertEqual(len(out_dict_with_video[self.videos_input_name]), 1)
|
||||
self.assertEqual(len(out_dict_with_video[self.videos_input_name][0]), 300)
|
||||
|
||||
# Load with `video_fps` and `num_frames` args, should raise an error
|
||||
# Load with `fps` and `num_frames` args, should raise an error
|
||||
with self.assertRaises(ValueError):
|
||||
out_dict_with_video = processor.apply_chat_template(
|
||||
messages,
|
||||
add_generation_prompt=True,
|
||||
tokenize=True,
|
||||
return_dict=True,
|
||||
video_fps=video_fps,
|
||||
fps=fps,
|
||||
num_frames=num_frames,
|
||||
)
|
||||
|
||||
|
||||
Reference in New Issue
Block a user