Factorized Learning for Temporally Grounded Video-Language Models
Paper • 2512.24097 • Published • 7
Dataset Preview
The full dataset viewer is not available (click to read why). Only showing a preview of the rows.
The dataset generation failed because of a cast error
Error code: DatasetGenerationCastError
Exception: DatasetGenerationCastError
Message: An error occurred while generating the dataset
All the data files must have the same columns, but at some point there are 1 new columns ({'tgt'}) and 3 missing columns ({'o', 'src', 'p'}).
This happened while the json dataset builder was generating data using
hf://datasets/wenzhengzeng/D2VLM-Dataset/ETBench/evi/tal_perception_test.json (at revision c3bf7b9da1cbaa282297563ecb9db9284346049c)
Please either edit the data files to have matching columns, or separate them into different configurations (see docs at https://hf.co/docs/hub/datasets-manual-configuration#multiple-configurations)
Traceback: Traceback (most recent call last):
File "/usr/local/lib/python3.12/site-packages/datasets/builder.py", line 1831, in _prepare_split_single
writer.write_table(table)
File "/usr/local/lib/python3.12/site-packages/datasets/arrow_writer.py", line 714, in write_table
pa_table = table_cast(pa_table, self._schema)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/usr/local/lib/python3.12/site-packages/datasets/table.py", line 2272, in table_cast
return cast_table_to_schema(table, schema)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/usr/local/lib/python3.12/site-packages/datasets/table.py", line 2218, in cast_table_to_schema
raise CastError(
datasets.table.CastError: Couldn't cast
version: int64
idx: int64
task: string
source: string
video: string
duration: double
tgt: list<item: list<item: double>>
child 0, item: list<item: double>
child 0, item: double
q: string
-- schema metadata --
pandas: '{"index_columns": [], "column_indexes": [], "columns": [{"name":' + 1018
to
{'version': Value('int64'), 'idx': Value('int64'), 'task': Value('string'), 'source': Value('string'), 'video': Value('string'), 'duration': Value('float64'), 'src': List(Value('float64')), 'p': Value('int64'), 'o': List(Value('string')), 'q': Value('string')}
because column names don't match
During handling of the above exception, another exception occurred:
Traceback (most recent call last):
File "/src/services/worker/src/worker/job_runners/config/parquet_and_info.py", line 1339, in compute_config_parquet_and_info_response
parquet_operations = convert_to_parquet(builder)
^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/src/services/worker/src/worker/job_runners/config/parquet_and_info.py", line 972, in convert_to_parquet
builder.download_and_prepare(
File "/usr/local/lib/python3.12/site-packages/datasets/builder.py", line 894, in download_and_prepare
self._download_and_prepare(
File "/usr/local/lib/python3.12/site-packages/datasets/builder.py", line 970, in _download_and_prepare
self._prepare_split(split_generator, **prepare_split_kwargs)
File "/usr/local/lib/python3.12/site-packages/datasets/builder.py", line 1702, in _prepare_split
for job_id, done, content in self._prepare_split_single(
^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/usr/local/lib/python3.12/site-packages/datasets/builder.py", line 1833, in _prepare_split_single
raise DatasetGenerationCastError.from_cast_error(
datasets.exceptions.DatasetGenerationCastError: An error occurred while generating the dataset
All the data files must have the same columns, but at some point there are 1 new columns ({'tgt'}) and 3 missing columns ({'o', 'src', 'p'}).
This happened while the json dataset builder was generating data using
hf://datasets/wenzhengzeng/D2VLM-Dataset/ETBench/evi/tal_perception_test.json (at revision c3bf7b9da1cbaa282297563ecb9db9284346049c)
Please either edit the data files to have matching columns, or separate them into different configurations (see docs at https://hf.co/docs/hub/datasets-manual-configuration#multiple-configurations)Need help to make the dataset viewer work? Make sure to review how to configure the dataset viewer, and open a discussion for direct support.
version int64 | idx int64 | task string | source string | video string | duration float64 | src list | p int64 | o list | q string |
|---|---|---|---|---|---|---|---|---|---|
1 | 0 | rar | perception_test | perception_test/video_2167.mp4 | 23.667 | [
4.574
] | 3 | [
"pretending to drink from something",
"shuffling objects",
"pretending to pour from something into something",
"pretending to open something"
] | You are given a video containing a series of actions. Watch the video carefully and identify the action around <evi> by choosing from a set of options. The format of your response should be: 'Best Option: (your choice)'. For example: 'Best Option: (A)'. Now I give you the options: (A) pretending to drink from something... |
1 | 1 | rar | perception_test | perception_test/video_4197.mp4 | 35 | [
2.5460000000000003
] | 3 | [
"pouring something from something into something",
"covering something with something",
"pretending to plug something into something",
"shuffling objects"
] | You are given a video containing a series of actions. Watch the video carefully and identify the action around <evi> by choosing from a set of options. The format of your response should be: 'Best Option: (your choice)'. For example: 'Best Option: (A)'. Now I give you the options: (A) pouring something from something i... |
1 | 2 | rar | perception_test | perception_test/video_6111.mp4 | 21 | [
1.259
] | 0 | [
"pretending to write or draw on a piece of paper",
"showing something to the camera",
"putting something on top of something",
"trying to turn something on but not working"
] | You are given a video containing a series of actions. Watch the video carefully and identify the action around <evi> by choosing from a set of options. The format of your response should be: 'Best Option: (your choice)'. For example: 'Best Option: (C)'. Now I give you the options: (A) pretending to write or draw on a p... |
1 | 3 | rar | perception_test | perception_test/video_10440.mp4 | 23.5 | [
12.391
] | 2 | [
"putting something on top of something",
"moving object(s) around",
"launching something",
"pretending to plug something into something"
] | You are given a video containing a series of actions. Watch the video carefully and identify the action around <evi> by choosing from a set of options. The format of your response should be: 'Best Option: (your choice)'. For example: 'Best Option: (B)'. Now I give you the options: (A) putting something on top of someth... |
1 | 4 | rar | perception_test | perception_test/video_2903.mp4 | 26.333 | [
15.767
] | 1 | [
"putting something into something",
"mixing something with something",
"indicating a region with the hand",
"opening something"
] | You are given a video containing a series of actions. Watch the video carefully and identify the action around <evi> by choosing from a set of options. The format of your response should be: 'Best Option: (your choice)'. For example: 'Best Option: (B)'. Now I give you the options: (A) putting something into something (... |
1 | 5 | rar | perception_test | perception_test/video_4445.mp4 | 26.667 | [
23.351
] | 1 | [
"mixing something with something",
"turning something off",
"closing something",
"unplugging something from something"
] | You are given a video containing a series of actions. Watch the video carefully and identify the action around <evi> by choosing from a set of options. The format of your response should be: 'Best Option: (your choice)'. For example: 'Best Option: (B)'. Now I give you the options: (A) mixing something with something (B... |
1 | 6 | rar | perception_test | perception_test/video_1589.mp4 | 35 | [
33.092
] | 1 | [
"pretending to tie something",
"mixing something with something",
"pouring something from something into something",
"holding something in a state"
] | You are given a video containing a series of actions. Watch the video carefully and identify the action around <evi> by choosing from a set of options. The format of your response should be: 'Best Option: (your choice)'. For example: 'Best Option: (A)'. Now I give you the options: (A) pretending to tie something (B) mi... |
1 | 7 | rar | perception_test | perception_test/video_1223.mp4 | 31.167 | [
19.519
] | 1 | [
"moving object(s) around",
"putting something on top of something",
"hitting something against/with something",
"pretending to iron something"
] | You are given a video containing a series of actions. Watch the video carefully and identify the action around <evi> by choosing from a set of options. The format of your response should be: 'Best Option: (your choice)'. For example: 'Best Option: (A)'. Now I give you the options: (A) moving object(s) around (B) puttin... |
1 | 8 | rar | perception_test | perception_test/video_6397.mp4 | 35 | [
26.342
] | 0 | [
"showing something to the camera",
"putting something on top of something",
"buttoning up a button",
"moving object(s) around"
] | You are given a video containing a series of actions. Watch the video carefully and identify the action around <evi> by choosing from a set of options. The format of your response should be: 'Best Option: (your choice)'. For example: 'Best Option: (D)'. Now I give you the options: (A) showing something to the camera (B... |
1 | 9 | rar | perception_test | perception_test/video_8609.mp4 | 22.333 | [
14.971
] | 0 | [
"rotating something",
"covering something with something",
"tying something",
"opening something"
] | You are given a video containing a series of actions. Watch the video carefully and identify the action around <evi> by choosing from a set of options. The format of your response should be: 'Best Option: (your choice)'. For example: 'Best Option: (A)'. Now I give you the options: (A) rotating something (B) covering so... |
1 | 10 | rar | perception_test | perception_test/video_2565.mp4 | 35 | [
3.185
] | 0 | [
"shuffling objects",
"opening something",
"mixing something with something",
"pretending to drink from something"
] | You are given a video containing a series of actions. Watch the video carefully and identify the action around <evi> by choosing from a set of options. The format of your response should be: 'Best Option: (your choice)'. For example: 'Best Option: (A)'. Now I give you the options: (A) shuffling objects (B) opening some... |
1 | 11 | rar | perception_test | perception_test/video_9456.mp4 | 30.5 | [
3.337
] | 3 | [
"showing something to the camera",
"clapping hands",
"putting something on top of something",
"writing or drawing something on a piece of paper"
] | You are given a video containing a series of actions. Watch the video carefully and identify the action around <evi> by choosing from a set of options. The format of your response should be: 'Best Option: (your choice)'. For example: 'Best Option: (A)'. Now I give you the options: (A) showing something to the camera (B... |
1 | 12 | rar | perception_test | perception_test/video_10174.mp4 | 24.167 | [
6.526
] | 3 | [
"taking something out of something",
"holding something in a state",
"opening something",
"pouring something from something into something"
] | You are given a video containing a series of actions. Watch the video carefully and identify the action around <evi> by choosing from a set of options. The format of your response should be: 'Best Option: (your choice)'. For example: 'Best Option: (B)'. Now I give you the options: (A) taking something out of something ... |
1 | 13 | rar | perception_test | perception_test/video_9253.mp4 | 25.5 | [
3.6790000000000003
] | 3 | [
"pretending to clap hands",
"pouring something from something into something",
"holding something in a state",
"moving object(s) around"
] | You are given a video containing a series of actions. Watch the video carefully and identify the action around <evi> by choosing from a set of options. The format of your response should be: 'Best Option: (your choice)'. For example: 'Best Option: (B)'. Now I give you the options: (A) pretending to clap hands (B) pouri... |
1 | 14 | rar | perception_test | perception_test/video_1755.mp4 | 23.5 | [
8.853
] | 1 | [
"browsing through something",
"plugging something into something",
"snapping fingers",
"turning something on"
] | You are given a video containing a series of actions. Watch the video carefully and identify the action around <evi> by choosing from a set of options. The format of your response should be: 'Best Option: (your choice)'. For example: 'Best Option: (A)'. Now I give you the options: (A) browsing through something (B) plu... |
1 | 15 | rar | perception_test | perception_test/video_9246.mp4 | 34.167 | [
31.437
] | 3 | [
"pretending to pour from something into something",
"opening something",
"putting something on top of something",
"rotating something"
] | You are given a video containing a series of actions. Watch the video carefully and identify the action around <evi> by choosing from a set of options. The format of your response should be: 'Best Option: (your choice)'. For example: 'Best Option: (B)'. Now I give you the options: (A) pretending to pour from something ... |
1 | 16 | rar | perception_test | perception_test/video_1467.mp4 | 30 | [
25.278
] | 2 | [
"putting something on top of something",
"clicking something",
"moving object(s) around",
"indicating a region with the hand"
] | You are given a video containing a series of actions. Watch the video carefully and identify the action around <evi> by choosing from a set of options. The format of your response should be: 'Best Option: (your choice)'. For example: 'Best Option: (D)'. Now I give you the options: (A) putting something on top of someth... |
1 | 17 | rar | perception_test | perception_test/video_10527.mp4 | 22.833 | [
15.578
] | 2 | [
"pushing something",
"breaking something",
"mixing something with something",
"clapping hands"
] | You are given a video containing a series of actions. Watch the video carefully and identify the action around <evi> by choosing from a set of options. The format of your response should be: 'Best Option: (your choice)'. For example: 'Best Option: (B)'. Now I give you the options: (A) pushing something (B) breaking som... |
1 | 18 | rar | perception_test | perception_test/video_6948.mp4 | 34.833 | [
16.438
] | 1 | [
"moving object(s) around",
"opening something",
"turning something off",
"rotating something"
] | You are given a video containing a series of actions. Watch the video carefully and identify the action around <evi> by choosing from a set of options. The format of your response should be: 'Best Option: (your choice)'. For example: 'Best Option: (D)'. Now I give you the options: (A) moving object(s) around (B) openin... |
1 | 19 | rar | perception_test | perception_test/video_3996.mp4 | 35 | [
1.035
] | 1 | [
"breaking something",
"indicating a region with the hand",
"moving object(s) around",
"putting something on top of something"
] | You are given a video containing a series of actions. Watch the video carefully and identify the action around <evi> by choosing from a set of options. The format of your response should be: 'Best Option: (your choice)'. For example: 'Best Option: (C)'. Now I give you the options: (A) breaking something (B) indicating ... |
1 | 20 | rar | perception_test | perception_test/video_6129.mp4 | 35 | [
10.93
] | 1 | [
"lifting something and placing it back down",
"closing something",
"pretending to mix",
"mixing something with something"
] | You are given a video containing a series of actions. Watch the video carefully and identify the action around <evi> by choosing from a set of options. The format of your response should be: 'Best Option: (your choice)'. For example: 'Best Option: (A)'. Now I give you the options: (A) lifting something and placing it b... |
1 | 21 | rar | perception_test | perception_test/video_1722.mp4 | 23.833 | [
5.777
] | 2 | [
"clicking something",
"closing something",
"opening something",
"putting something on top of something"
] | You are given a video containing a series of actions. Watch the video carefully and identify the action around <evi> by choosing from a set of options. The format of your response should be: 'Best Option: (your choice)'. For example: 'Best Option: (B)'. Now I give you the options: (A) clicking something (B) closing som... |
1 | 22 | rar | perception_test | perception_test/video_10500.mp4 | 35 | [
10.035
] | 1 | [
"pretending to break something",
"moving object(s) around",
"launching something",
"holding something in a state"
] | You are given a video containing a series of actions. Watch the video carefully and identify the action around <evi> by choosing from a set of options. The format of your response should be: 'Best Option: (your choice)'. For example: 'Best Option: (D)'. Now I give you the options: (A) pretending to break something (B) ... |
1 | 23 | rar | perception_test | perception_test/video_3003.mp4 | 32.667 | [
27.199
] | 3 | [
"clicking something",
"showing something to the camera",
"folding something",
"putting something on top of something"
] | You are given a video containing a series of actions. Watch the video carefully and identify the action around <evi> by choosing from a set of options. The format of your response should be: 'Best Option: (your choice)'. For example: 'Best Option: (C)'. Now I give you the options: (A) clicking something (B) showing som... |
1 | 24 | rar | perception_test | perception_test/video_2809.mp4 | 35 | [
28.483
] | 3 | [
"browsing through something",
"closing something",
"opening something",
"moving object(s) around"
] | You are given a video containing a series of actions. Watch the video carefully and identify the action around <evi> by choosing from a set of options. The format of your response should be: 'Best Option: (your choice)'. For example: 'Best Option: (C)'. Now I give you the options: (A) browsing through something (B) clo... |
1 | 25 | rar | perception_test | perception_test/video_4247.mp4 | 27.833 | [
24.18
] | 3 | [
"browsing through something",
"turning something off",
"holding something in a state",
"turning something on"
] | You are given a video containing a series of actions. Watch the video carefully and identify the action around <evi> by choosing from a set of options. The format of your response should be: 'Best Option: (your choice)'. For example: 'Best Option: (B)'. Now I give you the options: (A) browsing through something (B) tur... |
1 | 26 | rar | perception_test | perception_test/video_5262.mp4 | 35 | [
12.605
] | 2 | [
"unfolding something",
"pretending to unplug something from something",
"pretending to iron something",
"holding something in a state"
] | You are given a video containing a series of actions. Watch the video carefully and identify the action around <evi> by choosing from a set of options. The format of your response should be: 'Best Option: (your choice)'. For example: 'Best Option: (B)'. Now I give you the options: (A) unfolding something (B) pretending... |
1 | 27 | rar | perception_test | perception_test/video_3780.mp4 | 23.833 | [
2.598
] | 0 | [
"opening something",
"drinking something from something",
"holding something in a state",
"closing something"
] | You are given a video containing a series of actions. Watch the video carefully and identify the action around <evi> by choosing from a set of options. The format of your response should be: 'Best Option: (your choice)'. For example: 'Best Option: (C)'. Now I give you the options: (A) opening something (B) drinking som... |
1 | 28 | rar | perception_test | perception_test/video_4885.mp4 | 28.167 | [
17.854
] | 3 | [
"pretending to tie something",
"putting something on top of something",
"moving object(s) around",
"launching something"
] | You are given a video containing a series of actions. Watch the video carefully and identify the action around <evi> by choosing from a set of options. The format of your response should be: 'Best Option: (your choice)'. For example: 'Best Option: (A)'. Now I give you the options: (A) pretending to tie something (B) pu... |
1 | 29 | rar | perception_test | perception_test/video_8928.mp4 | 30.167 | [
8.2
] | 0 | [
"writing or drawing something on a piece of paper",
"moving object(s) around",
"hitting something against/with something",
"showing something to the camera"
] | You are given a video containing a series of actions. Watch the video carefully and identify the action around <evi> by choosing from a set of options. The format of your response should be: 'Best Option: (your choice)'. For example: 'Best Option: (A)'. Now I give you the options: (A) writing or drawing something on a ... |
1 | 30 | rar | perception_test | perception_test/video_11412.mp4 | 25.667 | [
9.511
] | 3 | [
"trying to close something but something prevents that",
"pretending to put something into something",
"dropping something on top of something",
"putting something on top of something"
] | You are given a video containing a series of actions. Watch the video carefully and identify the action around <evi> by choosing from a set of options. The format of your response should be: 'Best Option: (your choice)'. For example: 'Best Option: (D)'. Now I give you the options: (A) trying to close something but some... |
1 | 31 | rar | perception_test | perception_test/video_7899.mp4 | 24.5 | [
6.037
] | 1 | [
"pretending to open something",
"pretending to pour from something into something",
"dropping something on top of something",
"pretending to drink from something"
] | You are given a video containing a series of actions. Watch the video carefully and identify the action around <evi> by choosing from a set of options. The format of your response should be: 'Best Option: (your choice)'. For example: 'Best Option: (A)'. Now I give you the options: (A) pretending to open something (B) p... |
1 | 32 | rar | perception_test | perception_test/video_7559.mp4 | 22.5 | [
1.911
] | 2 | [
"putting something on top of something",
"mixing something with something",
"indicating a region with the hand",
"moving object(s) around"
] | You are given a video containing a series of actions. Watch the video carefully and identify the action around <evi> by choosing from a set of options. The format of your response should be: 'Best Option: (your choice)'. For example: 'Best Option: (C)'. Now I give you the options: (A) putting something on top of someth... |
1 | 33 | rar | perception_test | perception_test/video_8400.mp4 | 21.833 | [
11.274000000000001
] | 3 | [
"opening something",
"putting something on top of something",
"jingling something",
"covering something with something"
] | You are given a video containing a series of actions. Watch the video carefully and identify the action around <evi> by choosing from a set of options. The format of your response should be: 'Best Option: (your choice)'. For example: 'Best Option: (C)'. Now I give you the options: (A) opening something (B) putting some... |
1 | 34 | rar | perception_test | perception_test/video_860.mp4 | 35 | [
20.342
] | 3 | [
"plugging something into something",
"indicating a region with the hand",
"pretending to write or draw on a piece of paper",
"writing or drawing something on a piece of paper"
] | You are given a video containing a series of actions. Watch the video carefully and identify the action around <evi> by choosing from a set of options. The format of your response should be: 'Best Option: (your choice)'. For example: 'Best Option: (C)'. Now I give you the options: (A) plugging something into something ... |
1 | 35 | rar | perception_test | perception_test/video_11241.mp4 | 30.5 | [
14.751
] | 3 | [
"unplugging something from something",
"jingling something",
"drinking something from something",
"mixing something with something"
] | You are given a video containing a series of actions. Watch the video carefully and identify the action around <evi> by choosing from a set of options. The format of your response should be: 'Best Option: (your choice)'. For example: 'Best Option: (A)'. Now I give you the options: (A) unplugging something from somethin... |
1 | 36 | rar | perception_test | perception_test/video_9603.mp4 | 24 | [
20.894
] | 1 | [
"launching something",
"moving object(s) around",
"clapping hands",
"putting something on top of something"
] | You are given a video containing a series of actions. Watch the video carefully and identify the action around <evi> by choosing from a set of options. The format of your response should be: 'Best Option: (your choice)'. For example: 'Best Option: (D)'. Now I give you the options: (A) launching something (B) moving obj... |
1 | 37 | rar | perception_test | perception_test/video_7661.mp4 | 28.5 | [
23.965
] | 2 | [
"jingling something",
"shuffling objects",
"mixing something with something",
"putting something into something"
] | You are given a video containing a series of actions. Watch the video carefully and identify the action around <evi> by choosing from a set of options. The format of your response should be: 'Best Option: (your choice)'. For example: 'Best Option: (A)'. Now I give you the options: (A) jingling something (B) shuffling o... |
1 | 38 | rar | perception_test | perception_test/video_6332.mp4 | 20.167 | [
16.017
] | 2 | [
"moving object(s) around",
"pulling something",
"putting something on top of something",
"holding something in a state"
] | You are given a video containing a series of actions. Watch the video carefully and identify the action around <evi> by choosing from a set of options. The format of your response should be: 'Best Option: (your choice)'. For example: 'Best Option: (D)'. Now I give you the options: (A) moving object(s) around (B) pullin... |
1 | 39 | rar | perception_test | perception_test/video_11396.mp4 | 27.333 | [
6.552
] | 1 | [
"unplugging something from something",
"rotating something",
"ironing something",
"moving object(s) around"
] | You are given a video containing a series of actions. Watch the video carefully and identify the action around <evi> by choosing from a set of options. The format of your response should be: 'Best Option: (your choice)'. For example: 'Best Option: (A)'. Now I give you the options: (A) unplugging something from somethin... |
1 | 40 | rar | perception_test | perception_test/video_524.mp4 | 27.333 | [
7.135
] | 3 | [
"hitting something against/with something",
"unplugging something from something",
"buttoning up a button",
"browsing through something"
] | You are given a video containing a series of actions. Watch the video carefully and identify the action around <evi> by choosing from a set of options. The format of your response should be: 'Best Option: (your choice)'. For example: 'Best Option: (C)'. Now I give you the options: (A) hitting something against/with som... |
1 | 41 | rar | perception_test | perception_test/video_8592.mp4 | 28.5 | [
24.092
] | 3 | [
"breaking something",
"holding something in a state",
"pretending to drink from something",
"opening something"
] | You are given a video containing a series of actions. Watch the video carefully and identify the action around <evi> by choosing from a set of options. The format of your response should be: 'Best Option: (your choice)'. For example: 'Best Option: (C)'. Now I give you the options: (A) breaking something (B) holding som... |
1 | 42 | rar | perception_test | perception_test/video_265.mp4 | 35 | [
31.726
] | 2 | [
"tying something",
"moving object(s) around",
"covering something with something",
"pouring something from something into something"
] | You are given a video containing a series of actions. Watch the video carefully and identify the action around <evi> by choosing from a set of options. The format of your response should be: 'Best Option: (your choice)'. For example: 'Best Option: (B)'. Now I give you the options: (A) tying something (B) moving object(... |
1 | 43 | rar | perception_test | perception_test/video_2384.mp4 | 32.667 | [
13.251
] | 2 | [
"showing something to the camera",
"pushing something",
"writing or drawing something on a piece of paper",
"putting something on top of something"
] | You are given a video containing a series of actions. Watch the video carefully and identify the action around <evi> by choosing from a set of options. The format of your response should be: 'Best Option: (your choice)'. For example: 'Best Option: (A)'. Now I give you the options: (A) showing something to the camera (B... |
1 | 44 | rar | perception_test | perception_test/video_8446.mp4 | 27.667 | [
9.188
] | 1 | [
"mixing something with something",
"putting something into something",
"clapping hands",
"browsing through something"
] | You are given a video containing a series of actions. Watch the video carefully and identify the action around <evi> by choosing from a set of options. The format of your response should be: 'Best Option: (your choice)'. For example: 'Best Option: (A)'. Now I give you the options: (A) mixing something with something (B... |
1 | 45 | rar | perception_test | perception_test/video_3813.mp4 | 30.667 | [
19.758
] | 3 | [
"moving object(s) around",
"hitting something against/with something",
"putting something on top of something",
"putting something into something"
] | You are given a video containing a series of actions. Watch the video carefully and identify the action around <evi> by choosing from a set of options. The format of your response should be: 'Best Option: (your choice)'. For example: 'Best Option: (B)'. Now I give you the options: (A) moving object(s) around (B) hittin... |
1 | 46 | rar | perception_test | perception_test/video_11506.mp4 | 35 | [
3.329
] | 3 | [
"mixing something with something",
"pretending to pour from something into something",
"breaking something",
"putting something into something"
] | You are given a video containing a series of actions. Watch the video carefully and identify the action around <evi> by choosing from a set of options. The format of your response should be: 'Best Option: (your choice)'. For example: 'Best Option: (C)'. Now I give you the options: (A) mixing something with something (B... |
1 | 47 | rar | perception_test | perception_test/video_1924.mp4 | 25.5 | [
8.086
] | 2 | [
"opening something",
"showing something to the camera",
"writing or drawing something on a piece of paper",
"putting something on top of something"
] | You are given a video containing a series of actions. Watch the video carefully and identify the action around <evi> by choosing from a set of options. The format of your response should be: 'Best Option: (your choice)'. For example: 'Best Option: (A)'. Now I give you the options: (A) opening something (B) showing some... |
1 | 48 | rar | perception_test | perception_test/video_8185.mp4 | 24.167 | [
15.606
] | 2 | [
"breaking something",
"putting something into something",
"trying to close something but something prevents that",
"opening something"
] | You are given a video containing a series of actions. Watch the video carefully and identify the action around <evi> by choosing from a set of options. The format of your response should be: 'Best Option: (your choice)'. For example: 'Best Option: (B)'. Now I give you the options: (A) breaking something (B) putting som... |
1 | 49 | rar | perception_test | perception_test/video_8772.mp4 | 34.833 | [
30.646
] | 2 | [
"shuffling objects",
"tearing something into pieces",
"mixing something with something",
"unplugging something from something"
] | You are given a video containing a series of actions. Watch the video carefully and identify the action around <evi> by choosing from a set of options. The format of your response should be: 'Best Option: (your choice)'. For example: 'Best Option: (A)'. Now I give you the options: (A) shuffling objects (B) tearing some... |
1 | 50 | rar | perception_test | perception_test/video_8174.mp4 | 23.333 | [
5.028
] | 3 | [
"putting something on top of something",
"writing or drawing something on a piece of paper",
"snapping fingers",
"holding something in a state"
] | You are given a video containing a series of actions. Watch the video carefully and identify the action around <evi> by choosing from a set of options. The format of your response should be: 'Best Option: (your choice)'. For example: 'Best Option: (D)'. Now I give you the options: (A) putting something on top of someth... |
1 | 51 | rar | perception_test | perception_test/video_1750.mp4 | 26.667 | [
20.893
] | 2 | [
"pretending to launch something",
"holding something in a state",
"showing something to the camera",
"pouring something from something into something"
] | You are given a video containing a series of actions. Watch the video carefully and identify the action around <evi> by choosing from a set of options. The format of your response should be: 'Best Option: (your choice)'. For example: 'Best Option: (C)'. Now I give you the options: (A) pretending to launch something (B)... |
1 | 52 | rar | perception_test | perception_test/video_2643.mp4 | 27.667 | [
20.547
] | 0 | [
"putting something on top of something",
"holding something in a state",
"mixing something with something",
"pretending to cover something"
] | You are given a video containing a series of actions. Watch the video carefully and identify the action around <evi> by choosing from a set of options. The format of your response should be: 'Best Option: (your choice)'. For example: 'Best Option: (D)'. Now I give you the options: (A) putting something on top of someth... |
1 | 53 | rar | perception_test | perception_test/video_9653.mp4 | 22.167 | [
13.342
] | 2 | [
"pretending to open something",
"putting something into something",
"pretending to pour from something into something",
"mixing something with something"
] | You are given a video containing a series of actions. Watch the video carefully and identify the action around <evi> by choosing from a set of options. The format of your response should be: 'Best Option: (your choice)'. For example: 'Best Option: (C)'. Now I give you the options: (A) pretending to open something (B) p... |
1 | 54 | rar | perception_test | perception_test/video_6126.mp4 | 29.333 | [
3.541
] | 3 | [
"taking something out of something",
"ironing something",
"clapping hands",
"putting something into something"
] | You are given a video containing a series of actions. Watch the video carefully and identify the action around <evi> by choosing from a set of options. The format of your response should be: 'Best Option: (your choice)'. For example: 'Best Option: (B)'. Now I give you the options: (A) taking something out of something ... |
1 | 55 | rar | perception_test | perception_test/video_9830.mp4 | 35 | [
25.908
] | 0 | [
"opening something",
"putting something into something",
"pretending to write or draw on a piece of paper",
"pouring something from something into something"
] | You are given a video containing a series of actions. Watch the video carefully and identify the action around <evi> by choosing from a set of options. The format of your response should be: 'Best Option: (your choice)'. For example: 'Best Option: (C)'. Now I give you the options: (A) opening something (B) putting some... |
1 | 56 | rar | perception_test | perception_test/video_4831.mp4 | 30.5 | [
5.77
] | 3 | [
"launching something",
"pretending to tie something",
"putting something on top of something",
"moving object(s) around"
] | You are given a video containing a series of actions. Watch the video carefully and identify the action around <evi> by choosing from a set of options. The format of your response should be: 'Best Option: (your choice)'. For example: 'Best Option: (C)'. Now I give you the options: (A) launching something (B) pretending... |
1 | 57 | rar | perception_test | perception_test/video_7892.mp4 | 22.167 | [
14.577
] | 1 | [
"pouring something from something into something",
"mixing something with something",
"jingling something",
"moving object(s) around"
] | You are given a video containing a series of actions. Watch the video carefully and identify the action around <evi> by choosing from a set of options. The format of your response should be: 'Best Option: (your choice)'. For example: 'Best Option: (B)'. Now I give you the options: (A) pouring something from something i... |
1 | 58 | rar | perception_test | perception_test/video_8542.mp4 | 28.667 | [
10.762
] | 2 | [
"pretending to write or draw on a piece of paper",
"opening something",
"covering something with something",
"putting something on top of something"
] | You are given a video containing a series of actions. Watch the video carefully and identify the action around <evi> by choosing from a set of options. The format of your response should be: 'Best Option: (your choice)'. For example: 'Best Option: (C)'. Now I give you the options: (A) pretending to write or draw on a p... |
1 | 59 | rar | perception_test | perception_test/video_1915.mp4 | 35 | [
9.529
] | 2 | [
"putting something into something",
"drinking something from something",
"holding something in a state",
"putting something on top of something"
] | You are given a video containing a series of actions. Watch the video carefully and identify the action around <evi> by choosing from a set of options. The format of your response should be: 'Best Option: (your choice)'. For example: 'Best Option: (A)'. Now I give you the options: (A) putting something into something (... |
1 | 60 | rar | perception_test | perception_test/video_9770.mp4 | 20.167 | [
7.457
] | 3 | [
"shuffling objects",
"pretending to mix",
"mixing something with something",
"breaking something"
] | You are given a video containing a series of actions. Watch the video carefully and identify the action around <evi> by choosing from a set of options. The format of your response should be: 'Best Option: (your choice)'. For example: 'Best Option: (B)'. Now I give you the options: (A) shuffling objects (B) pretending t... |
1 | 61 | rar | perception_test | perception_test/video_3683.mp4 | 35 | [
3.857
] | 0 | [
"putting something on top of something",
"other",
"putting something into something",
"moving object(s) around"
] | You are given a video containing a series of actions. Watch the video carefully and identify the action around <evi> by choosing from a set of options. The format of your response should be: 'Best Option: (your choice)'. For example: 'Best Option: (B)'. Now I give you the options: (A) putting something on top of someth... |
1 | 62 | rar | perception_test | perception_test/video_7301.mp4 | 29.833 | [
6.475
] | 0 | [
"putting something into something",
"drinking something from something",
"moving object(s) around",
"mixing something with something"
] | You are given a video containing a series of actions. Watch the video carefully and identify the action around <evi> by choosing from a set of options. The format of your response should be: 'Best Option: (your choice)'. For example: 'Best Option: (B)'. Now I give you the options: (A) putting something into something (... |
1 | 63 | rar | perception_test | perception_test/video_5172.mp4 | 32 | [
30.16
] | 1 | [
"hitting something against/with something",
"mixing something with something",
"pretending to break something",
"tearing something into pieces"
] | You are given a video containing a series of actions. Watch the video carefully and identify the action around <evi> by choosing from a set of options. The format of your response should be: 'Best Option: (your choice)'. For example: 'Best Option: (D)'. Now I give you the options: (A) hitting something against/with som... |
1 | 64 | rar | perception_test | perception_test/video_7077.mp4 | 23 | [
8.996
] | 2 | [
"covering something with something",
"opening something",
"writing or drawing something on a piece of paper",
"showing something to the camera"
] | You are given a video containing a series of actions. Watch the video carefully and identify the action around <evi> by choosing from a set of options. The format of your response should be: 'Best Option: (your choice)'. For example: 'Best Option: (D)'. Now I give you the options: (A) covering something with something ... |
1 | 65 | rar | perception_test | perception_test/video_281.mp4 | 26 | [
4.069
] | 1 | [
"drinking something from something",
"opening something",
"pouring something from something into something",
"pulling something"
] | You are given a video containing a series of actions. Watch the video carefully and identify the action around <evi> by choosing from a set of options. The format of your response should be: 'Best Option: (your choice)'. For example: 'Best Option: (C)'. Now I give you the options: (A) drinking something from something ... |
1 | 66 | rar | perception_test | perception_test/video_6795.mp4 | 28.5 | [
8.177
] | 2 | [
"mixing something with something",
"jingling something",
"putting something on top of something",
"lifting something and placing it back down"
] | You are given a video containing a series of actions. Watch the video carefully and identify the action around <evi> by choosing from a set of options. The format of your response should be: 'Best Option: (your choice)'. For example: 'Best Option: (B)'. Now I give you the options: (A) mixing something with something (B... |
1 | 67 | rar | perception_test | perception_test/video_6523.mp4 | 26 | [
13.357
] | 0 | [
"covering something with something",
"holding something in a state",
"showing something to the camera",
"pretending to drink from something"
] | You are given a video containing a series of actions. Watch the video carefully and identify the action around <evi> by choosing from a set of options. The format of your response should be: 'Best Option: (your choice)'. For example: 'Best Option: (C)'. Now I give you the options: (A) covering something with something ... |
1 | 68 | rar | perception_test | perception_test/video_939.mp4 | 35 | [
5.665
] | 3 | [
"shuffling objects",
"jingling something",
"lifting something and placing it back down",
"putting something on top of something"
] | You are given a video containing a series of actions. Watch the video carefully and identify the action around <evi> by choosing from a set of options. The format of your response should be: 'Best Option: (your choice)'. For example: 'Best Option: (A)'. Now I give you the options: (A) shuffling objects (B) jingling som... |
1 | 69 | rar | perception_test | perception_test/video_3078.mp4 | 21.667 | [
13.463
] | 0 | [
"turning something off",
"dunking something",
"closing something",
"turning something on"
] | You are given a video containing a series of actions. Watch the video carefully and identify the action around <evi> by choosing from a set of options. The format of your response should be: 'Best Option: (your choice)'. For example: 'Best Option: (B)'. Now I give you the options: (A) turning something off (B) dunking ... |
1 | 70 | rar | perception_test | perception_test/video_7491.mp4 | 32.667 | [
8.17
] | 3 | [
"dropping something on top of something",
"writing or drawing something on a piece of paper",
"closing something",
"taking something out of something"
] | You are given a video containing a series of actions. Watch the video carefully and identify the action around <evi> by choosing from a set of options. The format of your response should be: 'Best Option: (your choice)'. For example: 'Best Option: (A)'. Now I give you the options: (A) dropping something on top of somet... |
1 | 71 | rar | perception_test | perception_test/video_2748.mp4 | 26 | [
21.126
] | 2 | [
"putting something on top of something",
"moving object(s) around",
"showing something to the camera",
"pretending to iron something"
] | You are given a video containing a series of actions. Watch the video carefully and identify the action around <evi> by choosing from a set of options. The format of your response should be: 'Best Option: (your choice)'. For example: 'Best Option: (C)'. Now I give you the options: (A) putting something on top of someth... |
1 | 72 | rar | perception_test | perception_test/video_6352.mp4 | 35 | [
14.484
] | 0 | [
"folding something",
"crumpling something",
"juggling something",
"taking something out of something"
] | You are given a video containing a series of actions. Watch the video carefully and identify the action around <evi> by choosing from a set of options. The format of your response should be: 'Best Option: (your choice)'. For example: 'Best Option: (D)'. Now I give you the options: (A) folding something (B) crumpling so... |
1 | 73 | rar | perception_test | perception_test/video_7872.mp4 | 22.167 | [
4.61
] | 0 | [
"lifting something and placing it back down",
"indicating a region with the hand",
"covering something with something",
"jingling something"
] | You are given a video containing a series of actions. Watch the video carefully and identify the action around <evi> by choosing from a set of options. The format of your response should be: 'Best Option: (your choice)'. For example: 'Best Option: (C)'. Now I give you the options: (A) lifting something and placing it b... |
1 | 74 | rar | perception_test | perception_test/video_2713.mp4 | 35 | [
7.715
] | 2 | [
"clicking something",
"mixing something with something",
"tearing something into pieces",
"pouring something from something into something"
] | You are given a video containing a series of actions. Watch the video carefully and identify the action around <evi> by choosing from a set of options. The format of your response should be: 'Best Option: (your choice)'. For example: 'Best Option: (A)'. Now I give you the options: (A) clicking something (B) mixing some... |
1 | 75 | rar | perception_test | perception_test/video_3896.mp4 | 27.5 | [
1.948
] | 0 | [
"tearing something into pieces",
"pushing something",
"pretending to pour from something into something",
"putting something into something"
] | You are given a video containing a series of actions. Watch the video carefully and identify the action around <evi> by choosing from a set of options. The format of your response should be: 'Best Option: (your choice)'. For example: 'Best Option: (B)'. Now I give you the options: (A) tearing something into pieces (B) ... |
1 | 76 | rar | perception_test | perception_test/video_5085.mp4 | 35 | [
29.258
] | 1 | [
"indicating a region with the hand",
"putting something on top of something",
"moving object(s) around",
"holding something in a state"
] | You are given a video containing a series of actions. Watch the video carefully and identify the action around <evi> by choosing from a set of options. The format of your response should be: 'Best Option: (your choice)'. For example: 'Best Option: (C)'. Now I give you the options: (A) indicating a region with the hand ... |
1 | 77 | rar | perception_test | perception_test/video_5273.mp4 | 23.167 | [
7.15
] | 1 | [
"jingling something",
"unplugging something from something",
"turning something on",
"clicking something"
] | You are given a video containing a series of actions. Watch the video carefully and identify the action around <evi> by choosing from a set of options. The format of your response should be: 'Best Option: (your choice)'. For example: 'Best Option: (B)'. Now I give you the options: (A) jingling something (B) unplugging ... |
1 | 78 | rar | perception_test | perception_test/video_10309.mp4 | 21.333 | [
15.253
] | 0 | [
"putting something into something",
"opening something",
"closing something",
"pretending to write or draw on a piece of paper"
] | You are given a video containing a series of actions. Watch the video carefully and identify the action around <evi> by choosing from a set of options. The format of your response should be: 'Best Option: (your choice)'. For example: 'Best Option: (D)'. Now I give you the options: (A) putting something into something (... |
1 | 79 | rar | perception_test | perception_test/video_6985.mp4 | 22.667 | [
10.45
] | 3 | [
"putting something on top of something",
"indicating a region with the hand",
"pretending to cover something",
"putting something into something"
] | You are given a video containing a series of actions. Watch the video carefully and identify the action around <evi> by choosing from a set of options. The format of your response should be: 'Best Option: (your choice)'. For example: 'Best Option: (D)'. Now I give you the options: (A) putting something on top of someth... |
1 | 80 | rar | perception_test | perception_test/video_347.mp4 | 35 | [
32.717
] | 3 | [
"opening something",
"putting something on top of something",
"pushing something",
"launching something"
] | You are given a video containing a series of actions. Watch the video carefully and identify the action around <evi> by choosing from a set of options. The format of your response should be: 'Best Option: (your choice)'. For example: 'Best Option: (D)'. Now I give you the options: (A) opening something (B) putting some... |
1 | 81 | rar | perception_test | perception_test/video_11051.mp4 | 23.833 | [
10.061
] | 0 | [
"lifting something and placing it back down",
"opening something",
"pretending to break something",
"putting something into something"
] | You are given a video containing a series of actions. Watch the video carefully and identify the action around <evi> by choosing from a set of options. The format of your response should be: 'Best Option: (your choice)'. For example: 'Best Option: (B)'. Now I give you the options: (A) lifting something and placing it b... |
1 | 82 | rar | perception_test | perception_test/video_3517.mp4 | 35 | [
31.708
] | 0 | [
"taking something out of something",
"pretending to cover something",
"closing something",
"putting something into something"
] | You are given a video containing a series of actions. Watch the video carefully and identify the action around <evi> by choosing from a set of options. The format of your response should be: 'Best Option: (your choice)'. For example: 'Best Option: (C)'. Now I give you the options: (A) taking something out of something ... |
1 | 83 | rar | perception_test | perception_test/video_3044.mp4 | 21 | [
0.984
] | 0 | [
"opening something",
"showing something to the camera",
"moving object(s) around",
"pretending to write or draw on a piece of paper"
] | You are given a video containing a series of actions. Watch the video carefully and identify the action around <evi> by choosing from a set of options. The format of your response should be: 'Best Option: (your choice)'. For example: 'Best Option: (B)'. Now I give you the options: (A) opening something (B) showing some... |
1 | 84 | rar | perception_test | perception_test/video_10370.mp4 | 23.167 | [
6.2
] | 2 | [
"unplugging something from something",
"mixing something with something",
"breaking something",
"putting something into something"
] | You are given a video containing a series of actions. Watch the video carefully and identify the action around <evi> by choosing from a set of options. The format of your response should be: 'Best Option: (your choice)'. For example: 'Best Option: (B)'. Now I give you the options: (A) unplugging something from somethin... |
1 | 85 | rar | perception_test | perception_test/video_10997.mp4 | 34.833 | [
23.152
] | 0 | [
"lifting something and placing it back down",
"rotating something",
"turning something off",
"pretending to tear something into pieces"
] | You are given a video containing a series of actions. Watch the video carefully and identify the action around <evi> by choosing from a set of options. The format of your response should be: 'Best Option: (your choice)'. For example: 'Best Option: (C)'. Now I give you the options: (A) lifting something and placing it b... |
1 | 86 | rar | perception_test | perception_test/video_10423.mp4 | 35 | [
1.521
] | 3 | [
"closing something",
"pouring something from something into something",
"pretending to pour from something into something",
"cutting something with something"
] | You are given a video containing a series of actions. Watch the video carefully and identify the action around <evi> by choosing from a set of options. The format of your response should be: 'Best Option: (your choice)'. For example: 'Best Option: (D)'. Now I give you the options: (A) closing something (B) pouring some... |
1 | 87 | rar | perception_test | perception_test/video_3422.mp4 | 27.833 | [
11.765
] | 0 | [
"pouring something from something into something",
"launching something",
"covering something with something",
"putting something into something"
] | You are given a video containing a series of actions. Watch the video carefully and identify the action around <evi> by choosing from a set of options. The format of your response should be: 'Best Option: (your choice)'. For example: 'Best Option: (D)'. Now I give you the options: (A) pouring something from something i... |
1 | 88 | rar | perception_test | perception_test/video_7314.mp4 | 35 | [
27.28
] | 1 | [
"covering something with something",
"putting something into something",
"showing something to the camera",
"putting something on top of something"
] | You are given a video containing a series of actions. Watch the video carefully and identify the action around <evi> by choosing from a set of options. The format of your response should be: 'Best Option: (your choice)'. For example: 'Best Option: (C)'. Now I give you the options: (A) covering something with something ... |
1 | 89 | rar | perception_test | perception_test/video_4708.mp4 | 24.5 | [
9.654
] | 3 | [
"working on a computer",
"browsing through something",
"turning something on",
"holding something in a state"
] | You are given a video containing a series of actions. Watch the video carefully and identify the action around <evi> by choosing from a set of options. The format of your response should be: 'Best Option: (your choice)'. For example: 'Best Option: (B)'. Now I give you the options: (A) working on a computer (B) browsing... |
1 | 90 | rar | perception_test | perception_test/video_4808.mp4 | 35 | [
31.84
] | 1 | [
"turning something off",
"covering something with something",
"holding something in a state",
"putting something into something"
] | You are given a video containing a series of actions. Watch the video carefully and identify the action around <evi> by choosing from a set of options. The format of your response should be: 'Best Option: (your choice)'. For example: 'Best Option: (B)'. Now I give you the options: (A) turning something off (B) covering... |
1 | 91 | rar | perception_test | perception_test/video_2413.mp4 | 31.333 | [
9.009
] | 1 | [
"closing something",
"putting something on top of something",
"moving object(s) around",
"pretending to cover something"
] | You are given a video containing a series of actions. Watch the video carefully and identify the action around <evi> by choosing from a set of options. The format of your response should be: 'Best Option: (your choice)'. For example: 'Best Option: (D)'. Now I give you the options: (A) closing something (B) putting some... |
1 | 92 | rar | perception_test | perception_test/video_4054.mp4 | 35 | [
25.88
] | 3 | [
"covering something with something",
"moving object(s) around",
"folding something",
"showing something to the camera"
] | You are given a video containing a series of actions. Watch the video carefully and identify the action around <evi> by choosing from a set of options. The format of your response should be: 'Best Option: (your choice)'. For example: 'Best Option: (A)'. Now I give you the options: (A) covering something with something ... |
1 | 93 | rar | perception_test | perception_test/video_3839.mp4 | 24.667 | [
3.86
] | 3 | [
"cutting something with something",
"showing something to the camera",
"putting something into something",
"opening something"
] | You are given a video containing a series of actions. Watch the video carefully and identify the action around <evi> by choosing from a set of options. The format of your response should be: 'Best Option: (your choice)'. For example: 'Best Option: (A)'. Now I give you the options: (A) cutting something with something (... |
1 | 94 | rar | perception_test | perception_test/video_3548.mp4 | 27 | [
9.34
] | 3 | [
"opening something",
"pretending to drink from something",
"pretending to tie something",
"pretending to pour from something into something"
] | You are given a video containing a series of actions. Watch the video carefully and identify the action around <evi> by choosing from a set of options. The format of your response should be: 'Best Option: (your choice)'. For example: 'Best Option: (A)'. Now I give you the options: (A) opening something (B) pretending t... |
1 | 95 | rar | perception_test | perception_test/video_5325.mp4 | 23.167 | [
8.822
] | 3 | [
"writing or drawing something on a piece of paper",
"tearing something into pieces",
"moving object(s) around",
"showing something to the camera"
] | You are given a video containing a series of actions. Watch the video carefully and identify the action around <evi> by choosing from a set of options. The format of your response should be: 'Best Option: (your choice)'. For example: 'Best Option: (B)'. Now I give you the options: (A) writing or drawing something on a ... |
1 | 96 | rar | perception_test | perception_test/video_11199.mp4 | 20.333 | [
9.89
] | 1 | [
"pretending to write or draw on a piece of paper",
"showing something to the camera",
"holding something in a state",
"launching something"
] | You are given a video containing a series of actions. Watch the video carefully and identify the action around <evi> by choosing from a set of options. The format of your response should be: 'Best Option: (your choice)'. For example: 'Best Option: (B)'. Now I give you the options: (A) pretending to write or draw on a p... |
1 | 97 | rar | perception_test | perception_test/video_107.mp4 | 26 | [
15.776
] | 2 | [
"putting something into something",
"folding something",
"mixing something with something",
"pretending to put something into something"
] | You are given a video containing a series of actions. Watch the video carefully and identify the action around <evi> by choosing from a set of options. The format of your response should be: 'Best Option: (your choice)'. For example: 'Best Option: (D)'. Now I give you the options: (A) putting something into something (... |
1 | 98 | rar | perception_test | perception_test/video_1111.mp4 | 34.833 | [
25.665
] | 2 | [
"moving object(s) around",
"pretending to cover something",
"lifting something and placing it back down",
"pretending to pour from something into something"
] | You are given a video containing a series of actions. Watch the video carefully and identify the action around <evi> by choosing from a set of options. The format of your response should be: 'Best Option: (your choice)'. For example: 'Best Option: (A)'. Now I give you the options: (A) moving object(s) around (B) preten... |
1 | 99 | rar | perception_test | perception_test/video_2353.mp4 | 22.333 | [
19.123
] | 2 | [
"showing something to the camera",
"pretending to write or draw on a piece of paper",
"putting something on top of something",
"turning something on"
] | You are given a video containing a series of actions. Watch the video carefully and identify the action around <evi> by choosing from a set of options. The format of your response should be: 'Best Option: (your choice)'. For example: 'Best Option: (D)'. Now I give you the options: (A) showing something to the camera (B... |
End of preview.
Dataset used in D2VLM
Here we provided the instruction for dataset preparation. We provided the transformed annotation (or steps for transformation) tailor for D2VLM, due to the new training objective design.
| Dataset Name | Role | Source |
|---|---|---|
| E.T. Bench | Evaluation Benchmark | Data |
| Charades-STA | Evaluation Benchmark | Data, Annotation |
| Youcook2 | Evaluation Benchmark | Data, Annotation |
| E.T. Instruct | Finetuning data | Data |
E.T. Bench
Download and process the data following official document.
Use the annotations provided in this Hugging Face repo (i.e.,
D2VLM-Dataset/ETBench/evi).
Charades-STA
- Follow the provided link in the table above to download the raw videos and the annotations. The folder is organized as follows.
xx/charades
├─ Charades_v1 (Downloaded and extracted raw videos from Charades_v1.zip)
│ ├─ 001TG.mp4
│ ├─ 003WS.mp4
│ └─ ...
└─ test.caption_coco_format.json
- Run the post-processing scripts (in project repo). Remember to update paths in the .py files referenced by the script below.
bash other_benchmark_organize/charades/run.sh
Youcook2
- Follow the provided link in the table above to download the raw videos and the annotations. The folder is organized as follows.
xx/youcook2
├─ raw_videos (Downloaded and extracted raw videos from raw_videos.tar.gz)
│ ├─ testing
│ ├─ training
│ └─ validation
└─ val.caption_coco_format.json
- Run the post-processing scripts (in project repo). Remember to update paths in the .py files referenced by the script below.
bash other_benchmark_organize/youcook2/run.sh
E.T. Instruct
Download and process the data following official document.
Use the annotations provided in this Hugging Face repo.
D2VLM-Dataset/ET-Instruct/evi.json--> For supervised finetuning (SFT)./D2VLM-Dataset/ET-Instruct/FPO/tokenized_fpo_annotation.pt--> For FPO.
For FPO, we applied an optimization: the dispreferred sample is appended after the preferred (positive) sample within the same sequence, and we adjust the positional encodings and attention mask accordingly. This makes training more efficient than naively training paired samples as separate batch items. One may check the
fpo_anno_gen/run.shin the project repo to learn more.
Citation
If you find our work useful in your research, please consider citing our paper:
@inproceedings{d2vlm,
title={Factorized Learning for Temporally Grounded Video-Language Models},
author={Zeng, Wenzheng and Gao, Difei and Shou, Mike Zheng and Ng, Hwee Tou},
booktitle={Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV)},
year={2025},
pages={20683-20693}
}
- Downloads last month
- 13