LabInstruct: Benchmarking Situated Instructional Video Generation for Lab Procedures
Paper • 2610.14020 • Published • 17
Dataset Preview
The full dataset viewer is not available (click to read why). Only showing a preview of the rows.
The dataset generation failed
Error code: DatasetGenerationError
Exception: CastError
Message: Couldn't cast
task_id: string
split: string
domain: string
discipline: string
title: string
description: string
initial_image: string
scene: struct<objects: list<item: struct<id: string, name: string, contents: string, material: string>>, sp (... 57 chars omitted)
child 0, objects: list<item: struct<id: string, name: string, contents: string, material: string>>
child 0, item: struct<id: string, name: string, contents: string, material: string>
child 0, id: string
child 1, name: string
child 2, contents: string
child 3, material: string
child 1, spatial_relations: list<item: string>
child 0, item: string
child 2, environment: string
action_sequence: list<item: struct<step: int64, action: string, target_object: string, contact: string, state_transit (... 39 chars omitted)
child 0, item: struct<step: int64, action: string, target_object: string, contact: string, state_transition: struct (... 27 chars omitted)
child 0, step: int64
child 1, action: string
child 2, target_object: string
child 3, contact: string
child 4, state_transition: struct<from: string, to: string>
child 0, from: string
child 1, to: string
prompt_for_gen: string
difficulty_features: struct<number_of_objects: int64, number_of_action_steps: int64, requires_liquid_handling: bool, requ (... 247 chars omitted)
child 0, number_of_objects: int64
child 1, number_of_action_steps: int64
child 2, requires_liquid_handling: bool
child 3, requires_tool_manipulation: bool
child 4, requires_fine_motor_control: bool
child 5, requires_transparent_object_reasoning: bool
child 6, requires_color_tracking: bool
child 7, requires_volume_change_detection: bool
child 8, requires_spatial_memory: bool
child 9, requires_sequential_ordering: bool
viewpoint: string
source_spec: string
items: list<item: struct<qa_id: string, dimension: string, importance: string, question: string, question_z (... 11 chars omitted)
child 0, item: struct<qa_id: string, dimension: string, importance: string, question: string, question_zh: string>
child 0, qa_id: string
child 1, dimension: string
child 2, importance: string
child 3, question: string
child 4, question_zh: string
task_level: string
to
{'task_id': Value('string'), 'task_level': Value('string'), 'source_spec': Value('string'), 'items': List({'qa_id': Value('string'), 'dimension': Value('string'), 'importance': Value('string'), 'question': Value('string'), 'question_zh': Value('string')})}
because column names don't match
Traceback: Traceback (most recent call last):
File "/usr/local/lib/python3.14/site-packages/datasets/builder.py", line 1827, in _prepare_split_single
for key, table in generator:
^^^^^^^^^
File "/src/services/worker/src/worker/job_runners/config/parquet_and_info.py", line 613, in wrapped
for item in generator(*args, **kwargs):
~~~~~~~~~^^^^^^^^^^^^^^^^^
File "/usr/local/lib/python3.14/site-packages/datasets/packaged_modules/json/json.py", line 343, in _generate_tables
self._cast_table(pa_table, json_field_paths=json_field_paths),
~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/usr/local/lib/python3.14/site-packages/datasets/packaged_modules/json/json.py", line 132, in _cast_table
pa_table = table_cast(pa_table, self.info.features.arrow_schema)
File "/usr/local/lib/python3.14/site-packages/datasets/table.py", line 2378, in table_cast
return cast_table_to_schema(table, schema)
File "/usr/local/lib/python3.14/site-packages/datasets/table.py", line 2306, in cast_table_to_schema
raise CastError(
...<3 lines>...
)
datasets.table.CastError: Couldn't cast
task_id: string
split: string
domain: string
discipline: string
title: string
description: string
initial_image: string
scene: struct<objects: list<item: struct<id: string, name: string, contents: string, material: string>>, sp (... 57 chars omitted)
child 0, objects: list<item: struct<id: string, name: string, contents: string, material: string>>
child 0, item: struct<id: string, name: string, contents: string, material: string>
child 0, id: string
child 1, name: string
child 2, contents: string
child 3, material: string
child 1, spatial_relations: list<item: string>
child 0, item: string
child 2, environment: string
action_sequence: list<item: struct<step: int64, action: string, target_object: string, contact: string, state_transit (... 39 chars omitted)
child 0, item: struct<step: int64, action: string, target_object: string, contact: string, state_transition: struct (... 27 chars omitted)
child 0, step: int64
child 1, action: string
child 2, target_object: string
child 3, contact: string
child 4, state_transition: struct<from: string, to: string>
child 0, from: string
child 1, to: string
prompt_for_gen: string
difficulty_features: struct<number_of_objects: int64, number_of_action_steps: int64, requires_liquid_handling: bool, requ (... 247 chars omitted)
child 0, number_of_objects: int64
child 1, number_of_action_steps: int64
child 2, requires_liquid_handling: bool
child 3, requires_tool_manipulation: bool
child 4, requires_fine_motor_control: bool
child 5, requires_transparent_object_reasoning: bool
child 6, requires_color_tracking: bool
child 7, requires_volume_change_detection: bool
child 8, requires_spatial_memory: bool
child 9, requires_sequential_ordering: bool
viewpoint: string
source_spec: string
items: list<item: struct<qa_id: string, dimension: string, importance: string, question: string, question_z (... 11 chars omitted)
child 0, item: struct<qa_id: string, dimension: string, importance: string, question: string, question_zh: string>
child 0, qa_id: string
child 1, dimension: string
child 2, importance: string
child 3, question: string
child 4, question_zh: string
task_level: string
to
{'task_id': Value('string'), 'task_level': Value('string'), 'source_spec': Value('string'), 'items': List({'qa_id': Value('string'), 'dimension': Value('string'), 'importance': Value('string'), 'question': Value('string'), 'question_zh': Value('string')})}
because column names don't match
The above exception was the direct cause of the following exception:
Traceback (most recent call last):
File "/src/services/worker/src/worker/job_runners/config/parquet_and_info.py", line 1369, in compute_config_parquet_and_info_response
parquet_operations, partial, estimated_dataset_info = stream_convert_to_parquet(
~~~~~~~~~~~~~~~~~~~~~~~~~^
builder, max_dataset_size_bytes=max_dataset_size_bytes
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
)
^
File "/src/services/worker/src/worker/job_runners/config/parquet_and_info.py", line 948, in stream_convert_to_parquet
builder._prepare_split(split_generator=splits_generators[split], file_format="parquet")
~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/usr/local/lib/python3.14/site-packages/datasets/builder.py", line 1694, in _prepare_split
for job_id, done, content in self._prepare_split_single(
~~~~~~~~~~~~~~~~~~~~~~~~~~^
gen_kwargs=gen_kwargs, job_id=job_id, **_prepare_split_args
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
):
^
File "/usr/local/lib/python3.14/site-packages/datasets/builder.py", line 1879, in _prepare_split_single
raise DatasetGenerationError("An error occurred while generating the dataset") from e
datasets.exceptions.DatasetGenerationError: An error occurred while generating the datasetNeed help to make the dataset viewer work? Make sure to review how to configure the dataset viewer, and open a discussion for direct support.
task_id string | task_level string | source_spec string | items list |
|---|---|---|---|
001_L1_001_agronomy | L1 | 001_L1_001_agronomy_spec.json | [
{
"qa_id": "q01",
"dimension": "Scene Consistency",
"importance": "standard",
"question": "Does the camera viewpoint remain stable and consistent with a nearby observer, without abrupt pans, jumps, or scene changes, as the operator transfers the scalpel and scrapes the plant root?",
"question_zh... |
002_L1_002_agronomy | L1 | 002_L1_002_agronomy_spec.json | [
{
"qa_id": "q01",
"dimension": "Scene Consistency",
"importance": "standard",
"question": "Does the camera maintain a consistent nearby-observer viewpoint looking down at the petri dish and bottle throughout the pouring action?",
"question_zh": "在整个倾倒动作过程中，摄像机是否保持一致的附近观察者视角俯视培养皿和试剂瓶？"
},
{
... |
003_L1_003_agronomy | L1 | 003_L1_003_agronomy_spec.json | [
{
"qa_id": "q01",
"dimension": "Scene Consistency",
"importance": "standard",
"question": "Does the camera maintain the nearby observer viewpoint looking down at the blue laboratory bench throughout the procedure?",
"question_zh": "在整个过程中，摄像机是否保持俯视蓝色实验台的附近观察者视角？"
},
{
"qa_id": "q02",
... |
004_L1_004_agronomy | L1 | 004_L1_004_agronomy_spec.json | [
{
"qa_id": "q01",
"dimension": "Scene Consistency",
"importance": "standard",
"question": "Does the video maintain a consistent camera viewpoint looking down at the stainless steel bench from a nearby observer perspective throughout the procedure?",
"question_zh": "在整个过程中，视频是否从附近观察者视角保持一致的摄像机视角俯... |
005_L1_005_agronomy | L1 | 005_L1_005_agronomy_spec.json | [
{
"qa_id": "q01",
"dimension": "Scene Consistency",
"importance": "standard",
"question": "Does the camera maintain a nearby observer viewpoint looking down at the graduated cylinder without abrupt changes in angle or distance during the pouring action?",
"question_zh": "在倾倒动作过程中，摄像机是否保持俯视量筒的附近观... |
006_L1_006_agronomy | L1 | 006_L1_006_agronomy_spec.json | [
{
"qa_id": "q01",
"dimension": "Scene Consistency",
"importance": "standard",
"question": "Does the wooden bench and the black trays remain in a consistent arrangement throughout the extraction, without sudden changes in the surrounding laboratory scene?",
"question_zh": "在整个提取过程中，木质实验台和黑色托盘是否始终... |
007_L1_006_agronomy | L1 | 007_L1_006_agronomy_spec.json | [
{
"qa_id": "q01",
"dimension": "Scene Consistency",
"importance": "standard",
"question": "Does the camera maintain a consistent downward-looking view of the wooden bench from the same nearby observer position throughout the video?",
"question_zh": "在整个视频中，摄像机是否从同一附近观察者位置保持一致的俯视木质实验台的视角？"
},
... |
008_L1_009_agronomy | L1 | 008_L1_009_agronomy_spec.json | [
{
"qa_id": "q01",
"dimension": "Scene Consistency",
"importance": "standard",
"question": "Does the camera viewpoint remain a stable overhead view of the white bench and the glass microscope slide throughout the procedure?",
"question_zh": "在整个过程中，摄像机视角是否始终保持为白色实验台和玻璃载玻片的稳定俯视视角？"
},
{
"q... |
009_L1_009_agronomy | L1 | 009_L1_009_agronomy_spec.json | [
{
"qa_id": "q01",
"dimension": "Scene Consistency",
"importance": "standard",
"question": "Does the camera maintain a stable viewpoint looking down at the bench during and after the cutting motion?",
"question_zh": "在切割动作期间和之后，摄像机是否保持俯视实验台的稳定视角？"
},
{
"qa_id": "q02",
"dimension": "Ob... |
010_L1_009_agronomy | L1 | 010_L1_009_agronomy_spec.json | [
{
"qa_id": "q01",
"dimension": "Scene Consistency",
"importance": "standard",
"question": "Does the camera maintain a consistent overhead viewpoint of the petri dish on the white laboratory bench throughout the entire procedure, without abrupt changes in angle or position?",
"question_zh": "在整个过... |
011_L1_012_materials_science | L1 | 011_L1_012_materials_science_spec.json | [
{
"qa_id": "q01",
"dimension": "Scene Consistency",
"importance": "standard",
"question": "Does the video maintain a continuous observer viewpoint looking down at the same laboratory bench area throughout the entire procedure?",
"question_zh": "在整个过程中，视频是否保持俯视同一实验台区域的连续观察者视角？"
},
{
"qa_i... |
012_L1_012_materials_science | L1 | 012_L1_012_materials_science_spec.json | [
{
"qa_id": "q01",
"dimension": "Scene Consistency",
"importance": "standard",
"question": "Does the camera maintain a consistently looking-down viewpoint at the white laboratory bench from a nearby observer angle throughout the procedure?",
"question_zh": "在整个过程中，摄像机是否从附近观察者角度保持一致地俯视白色实验台的视角？"
... |
013_L1_014_materials_science | L1 | 013_L1_014_materials_science_spec.json | [
{
"qa_id": "q01",
"dimension": "Scene Consistency",
"importance": "standard",
"question": "Does the camera keep a consistent nearby observer viewpoint looking down at the white laboratory bench throughout the video?",
"question_zh": "在整个视频中，摄像机是否保持一致的附近观察者视角俯视白色实验台？"
},
{
"qa_id": "q02",... |
014_L1_015_materials_science | L1 | 014_L1_015_materials_science_spec.json | [
{
"qa_id": "q01",
"dimension": "Scene Consistency",
"importance": "standard",
"question": "Does the camera viewpoint remain stable and looking down at the dark laboratory bench throughout the pouring and knocking procedure?",
"question_zh": "在整个倾倒和敲击过程中，摄像机视角是否保持稳定并俯视深色实验台？"
},
{
"qa_id"... |
015_L1_015_materials_science | L1 | 015_L1_015_materials_science_spec.json | [
{
"qa_id": "q01",
"dimension": "Scene Consistency",
"importance": "standard",
"question": "Does the camera maintain the same nearby observer viewpoint looking down at the dark laboratory bench throughout the kneading and gathering process?",
"question_zh": "在整个揉捏和聚拢过程中，摄像机是否保持同一附近观察者视角俯视深色实验台？"
... |
016_L1_020_chemistry | L1 | 016_L1_020_chemistry_spec.json | [
{
"qa_id": "q01",
"dimension": "Scene Consistency",
"importance": "standard",
"question": "Does the camera maintain a consistent first-person viewpoint looking down at the blue laboratory bench throughout the entire procedure without shifting to a different perspective or background?",
"question... |
017_L1_021_chemistry | L1 | 017_L1_021_chemistry_spec.json | [
{
"qa_id": "q01",
"dimension": "Scene Consistency",
"importance": "standard",
"question": "Does the camera maintain a consistent first-person downward view of the blue bench, glass funnel, glass beaker, and glass dropper throughout the entire rinsing procedure?",
"question_zh": "在整个冲洗过程中，摄像机是否始终... |
018_L1_022_chemistry | L1 | 018_L1_022_chemistry_spec.json | [
{
"qa_id": "q01",
"dimension": "Scene Consistency",
"importance": "standard",
"question": "Does the camera maintain a steady first-person top-down view of the dark gray bench and the basin throughout the entire video?",
"question_zh": "在整个视频中，摄像机是否保持深灰色实验台和盆的稳定第一人称俯视视角？"
},
{
"qa_id": "q... |
019_L1_023_chemistry | L1 | 019_L1_023_chemistry_spec.json | [
{
"qa_id": "q01",
"dimension": "Scene Consistency",
"importance": "standard",
"question": "Does the camera continue to look down at the same dark laboratory bench from a first-person perspective throughout the bottle opening, cap placement, and aspiration?",
"question_zh": "在整个开瓶、放盖和吸取过程中，摄像机是否继... |
020_L1_024_chemistry | L1 | 020_L1_024_chemistry_spec.json | [
{
"qa_id": "q01",
"dimension": "Scene Consistency",
"importance": "standard",
"question": "Does the camera maintain a top-down view of the light blue bench and the reagent bottle for the entire clip?",
"question_zh": "在整个片段中，摄像机是否保持浅蓝色实验台和试剂瓶的俯视视角？"
},
{
"qa_id": "q02",
"dimension": ... |
021_L1_025_chemistry | L1 | 021_L1_025_chemistry_spec.json | [
{
"qa_id": "q01",
"dimension": "Scene Consistency",
"importance": "standard",
"question": "Does the camera viewpoint remain stable at the nearby observer position throughout the rinsing procedure?",
"question_zh": "在整个冲洗过程中，摄像机视角是否在附近观察者位置保持稳定？"
},
{
"qa_id": "q02",
"dimension": "Obj... |
022_L1_026_chemistry | L1 | 022_L1_026_chemistry_spec.json | [
{
"qa_id": "q01",
"dimension": "Scene Consistency",
"importance": "standard",
"question": "Does the camera maintain a stable top-down view of the dark gray bench without abrupt viewpoint shifts throughout the procedure?",
"question_zh": "在整个过程中，摄像机是否保持深灰色实验台的稳定俯视视角，视角不发生突变？"
},
{
"qa_id"... |
023_L1_027_chemistry | L1 | 023_L1_027_chemistry_spec.json | [
{
"qa_id": "q01",
"dimension": "Scene Consistency",
"importance": "standard",
"question": "During the scooping and pouring actions, does the camera maintain a stable overhead view of the same dark laboratory bench without shifting to a different viewpoint?",
"question_zh": "在舀取和倾倒动作期间，摄像机是否保持同一深... |
024_L1_028_chemistry | L1 | 024_L1_028_chemistry_spec.json | [
{
"qa_id": "q01",
"dimension": "Scene Consistency",
"importance": "standard",
"question": "Does the camera maintain a steady downward view of the dark bench, with the blue-cap sample vial remaining on the right side of the frame, throughout the cap removal and membrane peeling?",
"question_zh": ... |
025_L1_029_chemistry | L1 | 025_L1_029_chemistry_spec.json | [
{
"qa_id": "q01",
"dimension": "Scene Consistency",
"importance": "standard",
"question": "Do the blurred white and blue pipette tip racks in the background remain visually continuous and consistent throughout the volume adjustment?",
"question_zh": "在整个调量过程中，背景中模糊的白色和蓝色移液器吸头架是否在视觉上保持连续和一致？"
}... |
026_L1_029_chemistry | L1 | 026_L1_029_chemistry_spec.json | [
{
"qa_id": "q01",
"dimension": "Scene Consistency",
"importance": "standard",
"question": "Does the camera viewpoint remain steady from the nearby observer perspective throughout the video?",
"question_zh": "在整个视频中，摄像机视角是否从附近观察者视角保持稳定？"
},
{
"qa_id": "q02",
"dimension": "Scene Consis... |
027_L1_031_chemistry | L1 | 027_L1_031_chemistry_spec.json | [
{
"qa_id": "q01",
"dimension": "Scene Consistency",
"importance": "standard",
"question": "Does the camera viewpoint remain stable and continuous, without cuts or abrupt shifts, throughout the procedure?",
"question_zh": "在整个操作过程中，摄像机视角是否保持稳定且连续，没有剪切或突然变化？"
},
{
"qa_id": "q02",
"dime... |
028_L1_031_chemistry | L1 | 028_L1_031_chemistry_spec.json | [
{
"qa_id": "q01",
"dimension": "Scene Consistency",
"importance": "standard",
"question": "Does the camera maintain a nearby observer viewpoint with the green bench surface and metal support stand continuously in view throughout the video?",
"question_zh": "在整个视频中，摄像机是否始终保持在近处观察者视角，绿色实验台面和金属支架持续... |
029_L1_031_chemistry | L1 | 029_L1_031_chemistry_spec.json | [
{
"qa_id": "q01",
"dimension": "Scene Consistency",
"importance": "standard",
"question": "Does the camera remain at a steady nearby observer viewpoint without sudden jumps or cuts while the glass funnel is being lowered and inserted into the copper funnel?",
"question_zh": "在玻璃漏斗被降下并插入铜漏斗的过程中，摄... |
030_L1_032_chemistry | L1 | 030_L1_032_chemistry_spec.json | [
{
"qa_id": "q01",
"dimension": "Scene Consistency",
"importance": "standard",
"question": "Does the camera maintain a stable observer viewpoint throughout the demonstration without sudden jumps or cuts?",
"question_zh": "在整个演示过程中，摄像机是否保持稳定的观察者视角，没有突然跳动或剪切？"
},
{
"qa_id": "q02",
"dime... |
031_L1_032_chemistry | L1 | 031_L1_032_chemistry_spec.json | [
{
"qa_id": "q01",
"dimension": "Scene Consistency",
"importance": "standard",
"question": "Does the green background and the metal retort stand remain continuously visible without abrupt scene cuts or camera jumps while the left hand brings and places the stopper?",
"question_zh": "在左手取来并放置瓶塞的过程... |
032_L1_032_chemistry | L1 | 032_L1_032_chemistry_spec.json | [
{
"qa_id": "q01",
"dimension": "Scene Consistency",
"importance": "standard",
"question": "Does the green background remain visually consistent (no sudden changes in color or pattern) throughout the video?",
"question_zh": "在整个视频中，绿色背景是否保持视觉上的一致（没有颜色或图案的突然变化）？"
},
{
"qa_id": "q02",
"... |
033_L1_032_chemistry | L1 | 033_L1_032_chemistry_spec.json | [
{
"qa_id": "q01",
"dimension": "Scene Consistency",
"importance": "standard",
"question": "Does the camera viewpoint remain stable at the nearby observer position throughout the video?",
"question_zh": "在整个视频中，摄像机视角是否保持在近处观察者位置保持稳定？"
},
{
"qa_id": "q02",
"dimension": "Object Consiste... |
034_L1_032_chemistry | L1 | 034_L1_032_chemistry_spec.json | [
{
"qa_id": "q01",
"dimension": "Scene Consistency",
"importance": "standard",
"question": "Does the camera viewpoint remain steady from the nearby observer position throughout the stopper-removal procedure?",
"question_zh": "在整个拔塞操作过程中，摄像机视角是否始终从近处观察者位置保持稳定？"
},
{
"qa_id": "q02",
"di... |
035_L1_033_chemistry | L1 | 035_L1_033_chemistry_spec.json | [
{
"qa_id": "q01",
"dimension": "Scene Consistency",
"importance": "standard",
"question": "Does the camera viewpoint remain steady from the nearby observer position throughout the video?",
"question_zh": "在整个视频中，摄像机视角是否始终从近处观察者位置保持稳定？"
},
{
"qa_id": "q02",
"dimension": "Object Consis... |
036_L1_033_chemistry | L1 | 036_L1_033_chemistry_spec.json | [
{
"qa_id": "q01",
"dimension": "Scene Consistency",
"importance": "standard",
"question": "Does the camera maintain a steady observer viewpoint with the laboratory bench, magnetic stirrer hotplate, and glass distillation apparatus remaining in the frame without sudden cuts or jumps?",
"question_... |
037_L1_033_chemistry | L1 | 037_L1_033_chemistry_spec.json | [
{
"qa_id": "q01",
"dimension": "Scene Consistency",
"importance": "standard",
"question": "Does the camera viewpoint remain stable as a nearby observer throughout the entire procedure?",
"question_zh": "在整个操作过程中，摄像机视角是否作为近处观察者始终保持稳定？"
},
{
"qa_id": "q02",
"dimension": "Scene Consiste... |
038_L1_033_chemistry | L1 | 038_L1_033_chemistry_spec.json | [
{
"qa_id": "q01",
"dimension": "Scene Consistency",
"importance": "standard",
"question": "Does the green laboratory bench and dark green background remain visually consistent and stable throughout the procedure?",
"question_zh": "在整个操作过程中，绿色实验台和深绿色背景是否保持视觉上的一致和稳定？"
},
{
"qa_id": "q02",
... |
039_L1_034_chemistry | L1 | 039_L1_034_chemistry_spec.json | [
{
"qa_id": "q01",
"dimension": "Scene Consistency",
"importance": "standard",
"question": "Does the video maintain a stable camera viewpoint from the nearby observer position without abrupt shifts or rotations?",
"question_zh": "视频是否从近处观察者位置保持稳定的摄像机视角，没有突然的移动或旋转？"
},
{
"qa_id": "q02",
... |
040_L1_034_chemistry | L1 | 040_L1_034_chemistry_spec.json | [
{
"qa_id": "q01",
"dimension": "Scene Consistency",
"importance": "standard",
"question": "Does the camera maintain a stable nearby observer viewpoint throughout the video without panning, zooming, or cutting?",
"question_zh": "在整个视频中，摄像机是否保持稳定的近处观察者视角，没有摇摄、变焦或剪切？"
},
{
"qa_id": "q02",
... |
041_L1_034_chemistry | L1 | 041_L1_034_chemistry_spec.json | [
{
"qa_id": "q01",
"dimension": "Scene Consistency",
"importance": "supplementary",
"question": "Does the camera viewpoint remain fixed throughout the procedure as the extractor is lifted and rotated?",
"question_zh": "在提取器被提起并旋转的整个操作过程中，摄像机视角是否保持固定？"
},
{
"qa_id": "q02",
"dimension":... |
042_L1_037_biology | L1 | 042_L1_037_biology_spec.json | [
{
"qa_id": "q01",
"dimension": "Scene Consistency",
"importance": "standard",
"question": "Does the camera maintain a stable first-person viewpoint looking down at the black bench and the 6-well plate throughout the video without any sudden shifts in angle or position?",
"question_zh": "在整个视频中，摄... |
043_L1_037_biology | L1 | 043_L1_037_biology_spec.json | [
{
"qa_id": "q01",
"dimension": "Scene Consistency",
"importance": "standard",
"question": "Does the camera maintain a consistent first-person viewpoint looking down at the black laboratory bench throughout the procedure?",
"question_zh": "在整个操作过程中，摄像机是否保持一致的第一人称视角，俯视黑色实验台？"
},
{
"qa_id":... |
044_L1_043_biology | L1 | 044_L1_043_biology_spec.json | [
{
"qa_id": "q01",
"dimension": "Scene Consistency",
"importance": "standard",
"question": "Does the camera maintain a consistent top-down view of the black bench and the blue tip box throughout the video?",
"question_zh": "在整个视频中，摄像机是否保持一致的俯视视角，观察黑色实验台和蓝色移液器吸头盒？"
},
{
"qa_id": "q02",
... |
045_L1_043_biology | L1 | 045_L1_043_biology_spec.json | [
{
"qa_id": "q01",
"dimension": "Scene Consistency",
"importance": "standard",
"question": "Does the camera maintain a steady downward first-person view of the bench without sudden jumps or shifts in perspective throughout the procedure?",
"question_zh": "在整个操作过程中，摄像机是否保持稳定的向下第一人称视角观察实验台，没有透视上的突然... |
046_L1_044_biology | L1 | 046_L1_044_biology_spec.json | [
{
"qa_id": "q01",
"dimension": "Scene Consistency",
"importance": "standard",
"question": "Does the video maintain a consistent downward viewpoint of the black laboratory bench throughout the aspiration procedure?",
"question_zh": "在整个吸取操作过程中，视频是否保持一致的向下视角，观察黑色实验台？"
},
{
"qa_id": "q02",
... |
047_L1_045_biology | L1 | 047_L1_045_biology_spec.json | [
{
"qa_id": "q01",
"dimension": "Scene Consistency",
"importance": "standard",
"question": "Does the video maintain a continuous first-person viewpoint looking down at the black laboratory bench, without sudden cuts or background changes, during the dispensing procedure?",
"question_zh": "在整个分液操作... |
End of preview.
YAML Metadata Warning:empty or missing yaml metadata in repo card
Check out the documentation for more information.
LabInstruct: Benchmarking Situated Instructional Video Generation for Lab Procedures
🌐 Homepage | 👉 Dataset | 📄 Paper | 💻 Code | 🏆 Leaderboard
ℹ️ This repository only hosts the LabInstruct annotation data: structured annotations and source-video links only. Videos, clips, and first frames are not included here; rebuild them with the LabInstruct code repository.
📢 Updates
- [2026-09] LabInstruct is online!
📝 TODO
- Release the annotation data: 204 task specifications, 204 QA checklists, 81 source clip-boundary annotations, and source-video links.
- Release the code: data preparation, video generation, and evaluation harness.
📑 Table of Contents
- 📜 Abstract
- 🌟 Project Overview
- 🚀 Setup
- 🗂 Repository Structure
- 📁 Data Files
- 🏆 Leaderboard
- 🎓 BibTeX
- 📧 Contact
- 🙏 Acknowledgements
📜 Abstract
Self-driving laboratories (SDLs) aim to automate the full experimental loop, from scientific decision-making to physical execution. Ideally, AI-generated plans could be carried out directly by robotic systems, but reliable automation remains difficult in complex, open-world laboratory environments, where experiments often involve fine-grained manipulation, long-horizon procedures, and substantial variation across tasks and setups. Humans therefore remain an important execution interface between AI-generated plans and physical experiments, creating a need for clear and effective human-facing experimental guidance. Because laboratory procedures are inherently visual, spatial, and dynamic, video is particularly well suited to communicating apparatus configurations, manipulation actions, temporal dependencies, and state changes. Recent advances in video generation now make it possible to synthesize experimental demonstrations directly from an initial workspace image and a natural-language instruction. However, whether such models can reliably communicate real laboratory procedures has not been systematically studied. We introduce LabInstruct, a benchmark for situated instructional video generation in real laboratories. LabInstruct contains 204 tasks across 5 scientific disciplines, with real reference executions and structured annotations of objects, actions, contacts, and state transitions. Evaluating 8 frontier image-to-video models, we find that visually plausible generations frequently remain procedurally incorrect, revealing a substantial gap between visual realism and the reliability required for experimental instruction.
🌟 Project Overview
Figure 1. LabInstruct evaluates generated laboratory videos for procedural correctness beyond visual plausibility.
🚀 Setup
ℹ️ The setup and usage below refer to the LabInstruct code repository. This repository contains only the files under
data/(see Data Files), the README figures, and the license;bench/andscripts/live in the code repository.
1. Environment setup
Python 3.10 or newer is required. Install ffmpeg and ffprobe for media reconstruction and video evaluation.
python -m venv .venv
source .venv/bin/activate
pip install -e .
Optional dependencies:
pip install -e ".[diffusers]"
pip install -e ".[qa]"
2. Prepare benchmark data
Place legally obtained source videos under data/source_videos/, using the source_video_id names in data/video_sources.csv.
python scripts/prepare_data.py --check
python scripts/prepare_data.py --clips --first-frames
python scripts/specs_to_tasks.py
3. Generate videos
Configure model environments and checkpoints in bench/models.yaml, then run:
python -m bench.cli gen \
--run exp1 \
--models wan2.2,ltx2.3,minimax-h3 \
--gpus 0,1,2,3
4. Evaluate videos
export JUDGE_API_BASE_URL="https://your-endpoint.example/v1"
export GPT_API_TOKEN="your-token"
python scripts/judge_videos_gpt.py --model-name all --fps 4
🗂 Repository Structure
.
├── data/
│ ├── video_sources.csv source-video provenance and URLs
│ ├── specs/ structured task specifications (204 tasks)
│ ├── checklists/ frozen QA checklists (204 tasks)
│ ├── source_annotations/ source clip boundaries (81 source videos)
│ └── rules/ annotation and evaluation prompts
├── assets/ figures used by this README
└── LICENSE Apache License 2.0
This repository releases links and annotations only. The third-party source videos, extracted clips, and first frames are not included. Download the source videos from the URLs in
data/video_sources.csvunder their original terms, then rebuild clips and first frames withscripts/prepare_data.pyin the LabInstruct code repository.
📁 Data Files
This section explains what every file under data/ is for. All annotation files are UTF-8 JSON or plain text; data/video_sources.csv is the only file that references the source videos.
data/video_sources.csv — source-video provenance
One row per source video (81 in total: 61 bilibili, 15 FineBio, 3 ExpVid, 2 YouTube). Columns:
| Column | Description |
|---|---|
source_video_id |
3-digit source video id (e.g. 001). Matches the annotation file in data/source_annotations/ and the source_video_id names used by scripts/prepare_data.py in the code repository. |
source_url |
Link to the original video. Use it to download the source video legally under its original terms; the video itself is not in this repository. |
platform |
Hosting platform / source dataset of the video. |
n_task_clips |
Number of task clips cut from this source video (sums to 204 across all rows). |
note |
Free-form notes. |
data/specs/ — task specifications (204 JSON files)
One file per benchmark task, named {index}_{split}_{source_id}_{discipline}_spec.json (e.g. 001_L1_001_agronomy_spec.json). These are the core structured annotations of each laboratory procedure:
| Field | Description |
|---|---|
task_id |
Unique task id, identical to the file stem. |
split |
Task level: L1 (101 tasks) = a single atomic step (1–3 closely related actions); L2 (103 tasks) = a complete short subprocedure with an ordered action sequence. |
domain |
Procedure domain, one or more of solid_handling, liquid_handling, measurement, instrument_operations, container_operations, heating_cooling, filtration_separation (multiple values joined by ` |
discipline |
Scientific discipline: agronomy, biology, chemistry, materials_science, or physics. |
title / description |
Short title and a plain-language description of the procedure. |
initial_image |
Relative path to the task's first frame, which serves as the image input for video generation. The frame itself is not included; rebuild it with prepare_data.py --first-frames in the code repository. |
scene |
Annotated scene: objects (each with id, name, contents, material), spatial_relations (natural-language spatial descriptions), and environment (e.g. indoor_laboratory). |
action_sequence |
Ordered action steps; each step has action, target_object, contact (e.g. hand_tool, tool_object), and state_transition (from / to states). |
prompt_for_gen |
The natural-language instruction used as the text prompt for video generation. |
difficulty_features |
Quantitative difficulty flags: number_of_objects, number_of_action_steps, and boolean requirements (requires_liquid_handling, requires_tool_manipulation, requires_fine_motor_control, requires_transparent_object_reasoning, requires_color_tracking, requires_volume_change_detection, requires_spatial_memory, requires_sequential_ordering). |
viewpoint |
Recorded viewpoint: first_person (67 tasks) or nearby_observer (137 tasks). |
data/checklists/ — QA checklists (204 JSON files)
One file per task, named {task_id}_qa.json. Each checklist is a frozen set of yes/no questions for judging whether a generated video executes the task correctly. The evaluation prompt data/rules/vlm_judge.txt consumes these checklists, and the same questions are what raters answered in the human study.
| Field | Description |
|---|---|
task_id / task_level / source_spec |
The task this checklist belongs to and the spec file it was generated from. |
items[].qa_id |
Question id within the checklist. |
items[].dimension |
Evaluation dimension, one of six: Action Fidelity, Object Consistency, State Correctness, Physical Plausibility, Visual Safety, Scene Consistency. |
items[].importance |
Question weight: critical or standard. |
items[].question / items[].question_zh |
The question in English and in Chinese. |
Across the 204 checklists there are 3,229 questions in total: 1,299 Action Fidelity, 567 Object Consistency, 448 State Correctness, 371 Physical Plausibility, 305 Visual Safety, and 239 Scene Consistency.
data/source_annotations/ — source clip boundaries (81 CSV files)
One file per source video, named {source_video_id}.txt. Columns: start_sec, end_sec, level. Each row marks one task clip inside the source video: its start/end time in seconds and the task level (L1/L2) of that clip. scripts/prepare_data.py --clips in the code repository uses these boundaries to cut data/video_clips/ from the downloaded source videos. The clips themselves are not included.
data/rules/ — annotation and evaluation prompts (4 files)
Plain-text prompts that reproduce the annotation and evaluation pipeline:
| File | Purpose |
|---|---|
L1_vlm_annotation.txt |
VLM prompt for drafting an L1 spec (atomic-action annotation) from a video clip. |
L2_vlm_annotation.txt |
VLM prompt for drafting an L2 spec (short-subprocedure annotation) from a video clip. |
qa_generation.txt |
Prompt for generating the task-specific QA checklist from a spec. |
vlm_judge.txt |
Prompt for judging a generated video against the frozen checklist. |
These prompts are written to produce structured drafts for human verification.
🏆 Leaderboard
GPT-5.6 Sol is evaluated at 4 FPS; Human is the pooled overall human rating. Higher is better; † marks commercial models.
| Rank | Model | Type | Organization | GPT-5.6 Sol | Human | Tier |
|---|---|---|---|---|---|---|
| 1 | MiniMax H3 | Open weight | MiniMax | 46.3 | 44.3 | T0 |
| 2 | Seedance 2.0 † | Commercial | ByteDance | 45.1 | 43.0 | T0 |
| 3 | Wan 3.0 † | Commercial | Alibaba | 44.4 | 37.3 | T0 |
| 4 | Wan 2.2-I2V-A14B | Open weight | Alibaba | 25.9 | 15.4 | T1 |
| 5 | Cosmos3 Super | Open weight | NVIDIA | 22.8 | 11.5 | T1 |
| 6 | LTX 2.3 | Open weight | Lightricks | 22.1 | 3.0 | T2 |
| 7 | LingBot Video | Open weight | Robbyant | 19.5 | 10.5 | T2 |
| 8 | Cosmos3 Nano | Open weight | NVIDIA | 19.4 | 5.8 | T2 |
Model rankings shift slightly across evaluation views, evaluators, sampling rates, and generation seeds, so we also report a Tier to keep the comparison robust.
🎓 BibTeX
If you find our work helpful, please consider citing it:
@misc{fu2026labinstruct,
title = {LabInstruct: Benchmarking Situated Instructional Video Generation for Lab Procedures},
author = {Yuming Fu and Weijia Wu and Jing Chen and Jiahao Tang and Feifei Chen and Hongyu Zhu and Xin Jin and Alex Jinpeng Wang},
year = {2026},
eprint = {2610.14020},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2610.14020}
}
📧 Contact
For questions, please open an issue in this repository or email Yuming Fu at yumingfu@csu.edu.cn.
🙏 Acknowledgements
We thank the creators of the source videos and the authors of FineBio, ExpVid, and the evaluated video-generation models for their work.
- Downloads last month
- 399