Datasets:

THULab
/

Kovavavvavava_libero_spatial_9

Exception:    SplitsNotFoundError
Message:      The split names could not be parsed from the dataset config.
Traceback:    Traceback (most recent call last):
                File "tsfile/tsfile_py_cpp.pyx", line 567, in tsfile.tsfile_py_cpp.tsfile_reader_new_c
              tsfile.exceptions.FileOpenError: 28: 
              
              The above exception was the direct cause of the following exception:
              
              Traceback (most recent call last):
                File "/usr/local/lib/python3.14/site-packages/datasets/inspect.py", line 286, in get_dataset_config_info
                  for split_generator in builder._split_generators(
                                         ~~~~~~~~~~~~~~~~~~~~~~~~~^
                      StreamingDownloadManager(base_path=builder.base_path, download_config=download_config)
                      ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
                  )
                  ^
                File "/usr/local/lib/python3.14/site-packages/datasets/packaged_modules/tsfile/tsfile.py", line 271, in _split_generators
                  scan = self._scan_metadata(all_files)
                File "/usr/local/lib/python3.14/site-packages/datasets/packaged_modules/tsfile/tsfile.py", line 318, in _scan_metadata
                  with self._open_reader(file) as reader:
                       ~~~~~~~~~~~~~~~~~^^^^^^
                File "/usr/local/lib/python3.14/site-packages/datasets/packaged_modules/tsfile/tsfile.py", line 742, in _open_reader
                  return TsFileReader(file)
                File "tsfile/tsfile_reader.pyx", line 323, in tsfile.tsfile_reader.TsFileReaderPy.__init__
              SystemError: <class '_weakrefset.WeakSet'> returned a result with an exception set
              
              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/split_names.py", line 68, in compute_split_names_from_streaming_response
                  for split in get_dataset_split_names(
                               ~~~~~~~~~~~~~~~~~~~~~~~^
                      path=dataset,
                      ^^^^^^^^^^^^^
                      config_name=config,
                      ^^^^^^^^^^^^^^^^^^^
                      token=hf_token,
                      ^^^^^^^^^^^^^^^
                  )
                  ^
                File "/usr/local/lib/python3.14/site-packages/datasets/inspect.py", line 340, in get_dataset_split_names
                  info = get_dataset_config_info(
                      path,
                  ...<6 lines>...
                      **config_kwargs,
                  )
                File "/usr/local/lib/python3.14/site-packages/datasets/inspect.py", line 291, in get_dataset_config_info
                  raise SplitsNotFoundError("The split names could not be parsed from the dataset config.") from err
              datasets.inspect.SplitsNotFoundError: The split names could not be parsed from the dataset config.

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.

LIBERO Spatial 9 TsFile

This dataset is an Apache TsFile conversion of Kovavavvavava/libero_spatial_9 (https://huggingface.co/datasets/Kovavavvavava/libero_spatial_9), a LeRobot-style LIBERO Spatial demonstration dataset. It contains nine spatial robot-manipulation tasks, 450 episodes, and 55,139 frame rows sampled at 30 fps.

Modalities: Time-series. The converted repository contains numeric robot state, action, point-cloud, event, timing, episode/task tags, and mirrored source metadata. Camera videos remain in the original Hugging Face dataset.

Source Dataset and Authors

Original dataset: Kovavavvavava/libero_spatial_9 (https://huggingface.co/datasets/Kovavavvavava/libero_spatial_9)
Pinned source revision: a9f782bb0c1418f4c4391d843012d07f162e7f09 (https://huggingface.co/datasets/Kovavavvavava/libero_spatial_9/tree/a9f782bb0c1418f4c4391d843012d07f162e7f09)
Hugging Face repository uploader: Haotian Zhan (Kovavavvavava) (https://huggingface.co/Kovavavvavava)
Benchmark paper: LIBERO: Benchmarking Knowledge Transfer for Lifelong Robot Learning (https://arxiv.org/abs/2306.03310)
LIBERO paper authors: Bo Liu, Yifeng Zhu, Chongkai Gao, Yihao Feng, Qiang Liu, Yuke Zhu, and Peter Stone
License: the source repository does not declare a license; no license is inferred here
LeRobot codebase version: v2.1
Source frame layout: data/chunk-{episode_chunk:03d}/episode_{episode_index:06d}.parquet
Source video layout: videos/chunk-{episode_chunk:03d}/{video_key}/episode_{episode_index:06d}.mp4

Tasks and Scale

task_index	Task	Episodes	Rows (source metadata)
0	pick up the black bowl on the cookie box and place it on the plate	50	5,052
1	pick up the black bowl next to the ramekin and place it on the plate	50	6,707
2	pick up the black bowl between the plate and the ramekin and place it on the plate	50	5,068
3	pick up the black bowl on the wooden cabinet and place it on the plate	50	6,880
4	pick up the black bowl next to the plate and place it on the plate	50	5,963
5	pick up the black bowl next to the cookie box and place it on the plate	50	6,312
6	pick up the black bowl from table center and place it on the plate	50	5,882
7	pick up the black bowl in the top drawer of the wooden cabinet and place it on the plate	50	7,479
8	pick up the black bowl on the ramekin and place it on the plate	50	5,796

Each task has 50 episodes. The converted train split contains 55,139 rows across 450 episode devices.

Converted Files

data/libero_spatial_9_part_000.tsfile: episodes 0-112, 13,026 rows, 1.76 MiB
data/libero_spatial_9_part_001.tsfile: episodes 113-224, 13,684 rows, 1.85 MiB
data/libero_spatial_9_part_002.tsfile: episodes 225-337, 13,746 rows, 1.84 MiB
data/libero_spatial_9_part_003.tsfile: episodes 338-449, 14,683 rows, 1.92 MiB
Table in all files: libero_spatial_9
Total rows: 55,139
Total episodes: 450
Total TsFile size: 7,727,266 bytes (7.37 MiB)
Time precision: milliseconds
Metadata: meta/ is mirrored from the source, with meta/info.json rewritten to describe the TsFile schema, source revision, conversion mapping, four-shard layout, and video policy.

TsFile Schema

Time is an INT64 millisecond timestamp computed as round(timestamp * 1000). It restarts within each episode.

TAG columns:

episode_index
task_index

Scalar FIELD columns:

frame_index
sample_index
next_event_idx
observation_state_0
observation_state_1
observation_state_2
observation_state_3
observation_state_4
observation_state_5
observation_state_6
observation_state_7
action_0
action_1
action_2
action_3
action_4
action_5
action_6
observation_points_gripper_pcds_0
observation_points_gripper_pcds_1
observation_points_gripper_pcds_2
observation_points_gripper_pcds_3
observation_points_gripper_pcds_4
observation_points_gripper_pcds_5
observation_points_gripper_pcds_6
observation_points_gripper_pcds_7
observation_points_gripper_pcds_8
observation_points_gripper_pcds_9
observation_points_gripper_pcds_10
observation_points_gripper_pcds_11
observation_points_goal_gripper_pcds_0
observation_points_goal_gripper_pcds_1
observation_points_goal_gripper_pcds_2
observation_points_goal_gripper_pcds_3
observation_points_goal_gripper_pcds_4
observation_points_goal_gripper_pcds_5
observation_points_goal_gripper_pcds_6
observation_points_goal_gripper_pcds_7
observation_points_goal_gripper_pcds_8
observation_points_goal_gripper_pcds_9
observation_points_goal_gripper_pcds_10
observation_points_goal_gripper_pcds_11

Flattened FLOAT FIELD groups:

observation.state -> observation_state_0 ... observation_state_7
action -> action_0 ... action_6
observation.points.gripper_pcds -> observation_points_gripper_pcds_0 ... observation_points_gripper_pcds_11
observation.points.goal_gripper_pcds -> observation_points_goal_gripper_pcds_0 ... observation_points_goal_gripper_pcds_11

The source timestamp column is not retained as a separate field because it is exactly represented by Time / 1000 seconds. The source index column is renamed to sample_index.

Encoding and compression policy (applied to every shard):

FLOAT/DOUBLE fields use GORILLA encoding with LZ4 compression.
INT32/INT64 fields use TS_2DIFF encoding with LZ4 compression.
The Time column uses TS_2DIFF with LZ4.
BOOLEAN fields, if present in a source variant, use RLE with LZ4.
episode_index and task_index are stored as TsFile TAG columns (device/tag mechanism), with dictionary-like tag values rather than repeated numeric fields.

Conversion Notes

The shared config-driven lerobot converter was used. Conversion scripts are retained locally by the dataset maintainer and are not included in this repository.
All 450 source frame Parquet files are merged into one logical table and then packaged as four episode-range TsFile shards. All four shards use the same table name and schema; filter by episode_index and task_index to select an episode or task.
Fixed-width vectors are flattened into scalar TsFile fields. Full source prefixes are preserved and dot is replaced with underscore.
Both point-cloud columns are fixed 4x3 arrays in the source despite their variable-length feature declaration; each is flattened to 12 FLOAT fields.
No source rows, episodes, tasks, state/action dimensions, event values, or point-cloud coordinates are intentionally dropped. Only redundant timestamp is omitted after Time synthesis.

Videos

Videos are not duplicated in this converted repository. The pinned source has 1,800 frame-aligned MP4 files in four streams:

videos/chunk-000/observation.images.cam_libero.color/ (https://huggingface.co/datasets/Kovavavvavava/libero_spatial_9/tree/a9f782bb0c1418f4c4391d843012d07f162e7f09/videos/chunk-000/observation.images.cam_libero.color) - 450 MP4 files
videos/chunk-000/observation.images.cam_libero.wrist/ (https://huggingface.co/datasets/Kovavavvavava/libero_spatial_9/tree/a9f782bb0c1418f4c4391d843012d07f162e7f09/videos/chunk-000/observation.images.cam_libero.wrist) - 450 MP4 files
videos/chunk-000/observation.images.cam_libero.transformed_depth/ (https://huggingface.co/datasets/Kovavavvavava/libero_spatial_9/tree/a9f782bb0c1418f4c4391d843012d07f162e7f09/videos/chunk-000/observation.images.cam_libero.transformed_depth) - 450 MP4 files
videos/chunk-000/observation.images.cam_libero.goal_gripper_proj/ (https://huggingface.co/datasets/Kovavavvavava/libero_spatial_9/tree/a9f782bb0c1418f4c4391d843012d07f162e7f09/videos/chunk-000/observation.images.cam_libero.goal_gripper_proj) - 450 MP4 files

The numeric TsFile rows remain aligned with the original videos through episode_index, frame_index, and the source metadata under meta/episodes/.

Validation

All four TsFile shards were opened and fully scanned with the Apache TsFile Java reader. Each has one table named libero_spatial_9, 2 TAG columns, and 42 FIELD columns. Read-back row counts are exactly 13,026, 13,684, 13,746, and 14,683; their sum is 55,139, matching the staged Parquet and source metadata.

Minimal Read Example

from tsfile import TsFileReader

reader = TsFileReader("data/libero_spatial_9_part_000.tsfile")
table = reader.get_all_table_schemas()["libero_spatial_9"]
print(table)
with reader.query_table(
    "libero_spatial_9",
    ["episode_index", "task_index", "frame_index", "sample_index", "action_0", "observation_state_0"],
    batch_size=65536,
) as result:
    batch = result.read_arrow_batch()
    print(batch.to_pandas().head())
reader.close()

Citation

If you use the benchmark context, cite LIBERO (arXiv:2306.03310). The converted repository is derived from the Hugging Face snapshot listed above.

Downloads last month: 54

Total file size:

7.75 MB

Paper for THULab/Kovavavvavava_libero_spatial_9

LIBERO: Benchmarking Knowledge Transfer for Lifelong Robot Learning

Paper • 2306.03310 • Published Jun 5, 2023 • 3