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
schema: string
dataset_root_base: string
split_manifest: string
expected_split_counts: struct<train: int64, validation: int64, test: int64>
child 0, train: int64
child 1, validation: int64
child 2, test: int64
split_frame_counts: struct<train: int64, validation: int64, test: int64>
child 0, train: int64
child 1, validation: int64
child 2, test: int64
source_group_counts: struct<legacy_pretrain: struct<episodes: int64, frames: int64>, current_finetune: struct<episodes: i (... 193 chars omitted)
child 0, legacy_pretrain: struct<episodes: int64, frames: int64>
child 0, episodes: int64
child 1, frames: int64
child 1, current_finetune: struct<episodes: int64, frames: int64>
child 0, episodes: int64
child 1, frames: int64
child 2, normal_10mm: struct<episodes: int64, frames: int64>
child 0, episodes: int64
child 1, frames: int64
child 3, normal_20mm: struct<episodes: int64, frames: int64>
child 0, episodes: int64
child 1, frames: int64
child 4, recovery_20mm_from_probe: struct<episodes: int64, frames: int64>
child 0, episodes: int64
child 1, frames: int64
camera_reports: struct<base_top_cam: struct<repo_id: string, fps: int64, total_episodes: int64, total_frames: int64, (... 1428 chars omitted)
child 0, base_top_cam: struct<repo_id: string, fps: int64, total_episodes: int64, total_frames: int64, video_codec: string, (... 265 chars omitted)
child 0, repo_id: string
child 1, fps: int64
...
st: int64>
child 0, train: int64
child 1, validation: int64
child 2, test: int64
child 2, normal_10mm: struct<train: int64, validation: int64, test: int64>
child 0, train: int64
child 1, validation: int64
child 2, test: int64
child 3, normal_20mm: struct<train: int64, validation: int64, test: int64>
child 0, train: int64
child 1, validation: int64
child 2, test: int64
child 4, recovery_20mm_from_probe: struct<train: int64, validation: int64, test: int64>
child 0, train: int64
child 1, validation: int64
child 2, test: int64
episodes: list<item: struct<lerobot_episode_index: int64, source_group: string, source_episode: string, sample (... 148 chars omitted)
child 0, item: struct<lerobot_episode_index: int64, source_group: string, source_episode: string, sample_index: int (... 136 chars omitted)
child 0, lerobot_episode_index: int64
child 1, source_group: string
child 2, source_episode: string
child 3, sample_index: int64
child 4, hole: struct<sample_index: int64, grid_x: int64, grid_z: int64, repeat_index: int64, offset_xyz: list<item (... 10 chars omitted)
child 0, sample_index: int64
child 1, grid_x: int64
child 2, grid_z: int64
child 3, repeat_index: int64
child 4, offset_xyz: list<item: double>
child 0, item: double
child 5, split: string
to
{'schema_version': Value('string'), 'description': Value('string'), 'processed_input_root_templates': List(Value('string')), 'summary': {'ok': Value('bool'), 'episodes': Value('int64'), 'train': Value('int64'), 'validation': Value('int64'), 'test': Value('int64'), 'group_counts': {'legacy_pretrain': {'train': Value('int64'), 'validation': Value('int64'), 'test': Value('int64')}, 'current_finetune': {'train': Value('int64'), 'validation': Value('int64'), 'test': Value('int64')}, 'normal_10mm': {'train': Value('int64'), 'validation': Value('int64'), 'test': Value('int64')}, 'normal_20mm': {'train': Value('int64'), 'validation': Value('int64'), 'test': Value('int64')}, 'recovery_20mm_from_probe': {'train': Value('int64'), 'validation': Value('int64'), 'test': Value('int64')}}}, 'episode_indices': {'train': List(Value('int64')), 'validation': List(Value('int64')), 'test': List(Value('int64'))}, 'episodes': List({'lerobot_episode_index': Value('int64'), 'source_group': Value('string'), 'source_episode': Value('string'), 'sample_index': Value('int64'), 'hole': {'sample_index': Value('int64'), 'grid_x': Value('int64'), 'grid_z': Value('int64'), 'repeat_index': Value('int64'), 'offset_xyz': List(Value('float64'))}, 'split': 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
schema: string
dataset_root_base: string
split_manifest: string
expected_split_counts: struct<train: int64, validation: int64, test: int64>
child 0, train: int64
child 1, validation: int64
child 2, test: int64
split_frame_counts: struct<train: int64, validation: int64, test: int64>
child 0, train: int64
child 1, validation: int64
child 2, test: int64
source_group_counts: struct<legacy_pretrain: struct<episodes: int64, frames: int64>, current_finetune: struct<episodes: i (... 193 chars omitted)
child 0, legacy_pretrain: struct<episodes: int64, frames: int64>
child 0, episodes: int64
child 1, frames: int64
child 1, current_finetune: struct<episodes: int64, frames: int64>
child 0, episodes: int64
child 1, frames: int64
child 2, normal_10mm: struct<episodes: int64, frames: int64>
child 0, episodes: int64
child 1, frames: int64
child 3, normal_20mm: struct<episodes: int64, frames: int64>
child 0, episodes: int64
child 1, frames: int64
child 4, recovery_20mm_from_probe: struct<episodes: int64, frames: int64>
child 0, episodes: int64
child 1, frames: int64
camera_reports: struct<base_top_cam: struct<repo_id: string, fps: int64, total_episodes: int64, total_frames: int64, (... 1428 chars omitted)
child 0, base_top_cam: struct<repo_id: string, fps: int64, total_episodes: int64, total_frames: int64, video_codec: string, (... 265 chars omitted)
child 0, repo_id: string
child 1, fps: int64
...
st: int64>
child 0, train: int64
child 1, validation: int64
child 2, test: int64
child 2, normal_10mm: struct<train: int64, validation: int64, test: int64>
child 0, train: int64
child 1, validation: int64
child 2, test: int64
child 3, normal_20mm: struct<train: int64, validation: int64, test: int64>
child 0, train: int64
child 1, validation: int64
child 2, test: int64
child 4, recovery_20mm_from_probe: struct<train: int64, validation: int64, test: int64>
child 0, train: int64
child 1, validation: int64
child 2, test: int64
episodes: list<item: struct<lerobot_episode_index: int64, source_group: string, source_episode: string, sample (... 148 chars omitted)
child 0, item: struct<lerobot_episode_index: int64, source_group: string, source_episode: string, sample_index: int (... 136 chars omitted)
child 0, lerobot_episode_index: int64
child 1, source_group: string
child 2, source_episode: string
child 3, sample_index: int64
child 4, hole: struct<sample_index: int64, grid_x: int64, grid_z: int64, repeat_index: int64, offset_xyz: list<item (... 10 chars omitted)
child 0, sample_index: int64
child 1, grid_x: int64
child 2, grid_z: int64
child 3, repeat_index: int64
child 4, offset_xyz: list<item: double>
child 0, item: double
child 5, split: string
to
{'schema_version': Value('string'), 'description': Value('string'), 'processed_input_root_templates': List(Value('string')), 'summary': {'ok': Value('bool'), 'episodes': Value('int64'), 'train': Value('int64'), 'validation': Value('int64'), 'test': Value('int64'), 'group_counts': {'legacy_pretrain': {'train': Value('int64'), 'validation': Value('int64'), 'test': Value('int64')}, 'current_finetune': {'train': Value('int64'), 'validation': Value('int64'), 'test': Value('int64')}, 'normal_10mm': {'train': Value('int64'), 'validation': Value('int64'), 'test': Value('int64')}, 'normal_20mm': {'train': Value('int64'), 'validation': Value('int64'), 'test': Value('int64')}, 'recovery_20mm_from_probe': {'train': Value('int64'), 'validation': Value('int64'), 'test': Value('int64')}}}, 'episode_indices': {'train': List(Value('int64')), 'validation': List(Value('int64')), 'test': List(Value('int64'))}, 'episodes': List({'lerobot_episode_index': Value('int64'), 'source_group': Value('string'), 'source_episode': Value('string'), 'sample_index': Value('int64'), 'hole': {'sample_index': Value('int64'), 'grid_x': Value('int64'), 'grid_z': Value('int64'), 'repeat_index': Value('int64'), 'offset_xyz': List(Value('float64'))}, 'split': 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 1880, 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.
schema_version string | description string | processed_input_root_templates list | summary dict | episode_indices dict | episodes list |
|---|---|---|---|---|---|
act_training_v2_split_manifest_v1 | Fixed matched split for four ACT camera models. Historical 138 episodes are train-only; new random-hole data is 200/50/50. | [
"data/act_checkpoint_training_v1/legacy_pretrain/{camera}",
"data/act_checkpoint_training_v1/current_finetune/{camera}",
"data/act_random_hole_training_v1/normal_10mm/{camera}",
"data/act_random_hole_training_v1/normal_20mm/{camera}",
"data/act_random_hole_training_v1/recovery_20mm_from_probe/{camera}"
] | {
"ok": true,
"episodes": 438,
"train": 338,
"validation": 50,
"test": 50,
"group_counts": {
"legacy_pretrain": {
"train": 100,
"validation": 0,
"test": 0
},
"current_finetune": {
"train": 38,
"validation": 0,
"test": 0
},
"normal_10mm": {
"train... | {
"train": [
0,
1,
2,
3,
4,
5,
6,
7,
8,
9,
10,
11,
12,
13,
14,
15,
16,
17,
18,
19,
20,
21,
22,
23,
24,
25,
26,
27,
28,
29,
30,
31,
32,
33,
34,
35,
36,
37,
38,
... | [
{
"lerobot_episode_index": 0,
"source_group": "legacy_pretrain",
"source_episode": "20260701_131034_teleop_006",
"sample_index": null,
"hole": null,
"split": "train"
},
{
"lerobot_episode_index": 1,
"source_group": "legacy_pretrain",
"source_episode": "20260701_131116_teleop_... |
Peg-in-hole ACT-v2 multiview dataset
This repository contains 438 independent peg-in-hole trajectories. Camera-specific and format-specific copies of an episode are alternate representations of the same source trajectory and are not counted as additional trajectories.
Dataset organization
| Folder | Independent trajectories | Description |
|---|---|---|
01_legacy_pretrain_100 |
100 | Earlier human teleoperation demonstrations used for pretraining |
02_current_finetune_38 |
38 | Later human teleoperation demonstrations used for checkpoint fine-tuning |
03_automatic_trajectories_300 |
300 | Automatically generated normal-insertion and recovery demonstrations |
04_merged_lerobot_act_v2 |
438 per camera | Training-ready merged LeRobot exports of the same source trajectories |
manifests |
- | Fixed split, trimming, validation, and upload inventory metadata |
The source-level count is 100 + 38 + 300 = 438.
Automatic trajectory groups
| Folder | Count | Position range | Internal coverage |
|---|---|---|---|
01_normal_10mm_50 |
50 | approximately +/-10 mm | 5 x 5 position grid, 2 repeats per cell |
02_normal_20mm_100 |
100 | approximately +/-20 mm | 5 x 5 position grid, 4 repeats per cell |
03_recovery_20mm_from_probe_150 |
150 | approximately +/-20 mm | 5 x 5 position grid, 6 repeats per cell |
All 300 automatic trajectories were successfully executed; their collection manifests report zero rejected episodes.
The recovery demonstrations contain a controlled error, probe, withdrawal, realignment, and insertion sequence. Processed ACT targets begin at the first aligned automatic_stage_error_probe frame so that the deliberately injected error prefix is not used as an imitation target. Original raw HDF5 files are unchanged.
Internal folder structure
01_legacy_pretrain_100/
raw_multiview/
act_hdf5_two_view_v2/
base_top_cam/
cctv_cam/
wall_task_cam/
wall_side_cam/
02_current_finetune_38/
raw_multiview/
act_hdf5_two_view_v2/
base_top_cam/
cctv_cam/
wall_task_cam/
wall_side_cam/
03_automatic_trajectories_300/
01_normal_10mm_50/
raw_multiview/
act_hdf5_two_view_v2/<camera>/
02_normal_20mm_100/
raw_multiview/
act_hdf5_two_view_v2/<camera>/
03_recovery_20mm_from_probe_150/
raw_multiview/
act_hdf5_two_view_v2/<camera>/
04_merged_lerobot_act_v2/
base_top_cam/
cctv_cam/
wall_task_cam/
wall_side_cam/
manifests/
Every processed ACT file pairs the wrist view (ee_cam) with one of four environment views:
base_top_camcctv_camwall_task_camwall_side_cam
Fixed episode split
| Source group | Train | Validation | Test | Total |
|---|---|---|---|---|
| Legacy pretrain | 100 | 0 | 0 | 100 |
| Current fine-tune | 38 | 0 | 0 | 38 |
| Normal +/-10 mm | 25 | 13 | 12 | 50 |
| Normal +/-20 mm | 75 | 12 | 13 | 100 |
| Recovery +/-20 mm | 100 | 25 | 25 | 150 |
| Total | 338 | 50 | 50 | 438 |
The historical 138 episodes are train-only because earlier checkpoints have already seen them. Validation and test contain only held-out automatically generated episodes. All camera views use the same episode indices.
Representations
Raw multiview
Original MuJoCo trajectory HDF5 files plus per-episode provenance metadata and conversion or rerender reports.
ACT HDF5 two-view v2
Processed ACT inputs containing ee_cam, one environment camera, robot state, action, and force data. There are four camera-specific files per source episode.
Merged LeRobot ACT-v2
Each camera directory contains 438 episodes, 65,567 aligned frames at 30 fps, and the same fixed 338 / 50 / 50 episode split. The validation report is included in manifests/validation_report.json.
Manifests
act_training_v2_split_manifest.json: canonical source-to-index mapping and fixed split.legacy_checkpoint_trim_manifest.json: trimming record for the legacy human demonstrations.current_finetune_checkpoint_trim_manifest.json: trimming record for the 38 fine-tuning demonstrations.validation_report.json: final four-camera LeRobot validation report.upload_inventory.json: uploaded file counts and logical byte sizes by top-level folder.
Local machine paths in text metadata were replaced with placeholders before publication. Binary trajectory contents were not modified.
- Downloads last month
- 200