Dataset Viewer
The dataset viewer is not available for this subset.
Cannot get the split names for the config 'default' of the dataset.
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 71, 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.

LSY-lab Libero+ Goal Noise All Sawyer Failures (TsFile)

This repository contains the numeric time-series portion of the LSY-lab/libero_plus_goal_noise_all_sawyer_failures dataset converted from LeRobot Parquet to Apache TsFile. The source dataset was created with LeRobot and describes Sawyer robot demonstrations collected for the Libero+ goal-noise failure setting.

Source and attribution

TsFile contents

The converted file is data/libero_plus_goal_noise_all_sawyer_failures.tsfile (one merged table with 42,456 rows). Each episode_index + task_index pair is a TsFile table TAG/device identity; Time restarts at zero for each episode and is round(timestamp * 1000) in milliseconds.

Role Columns
Time Time (INT64, milliseconds)
TAG/device episode_index, task_index (INT64 source identifiers)
Integer FIELD frame_index, sample_index (INT64)
FLOAT FIELD Flattened state/action columns listed below

Vector columns retain their full source name (with dots replaced by underscores) and append an element index:

  • observation.state -> observation_state_0 ... observation_state_7
  • observation.states.ee_state -> observation_states_ee_state_0 ... _5
  • observation.states.joint_state -> observation_states_joint_state_0 ... _6
  • observation.states.gripper_state -> observation_states_gripper_state_0 ... _1
  • action -> action_0 ... action_6

The source timestamp column is intentionally dropped because it is represented by Time; index is retained as sample_index. No source rows are dropped. Video feature columns are omitted from TsFile because TsFile stores the numeric table only.

Encoding and size

The converter uses the compact profile requested for this dataset: FLOAT fields use GORILLA encoding, INT32/INT64 and Time use TS_2DIFF, and physical columns use LZ4 compression. TAGs are stored through the TsFile table/device mechanism. The resulting TsFile is approximately 5.26 MB versus the 7.74 MB source Parquet frame file.

Video location and alignment

Videos remain in the original Hugging Face repository and are not copied here. They are stored under https://huggingface.co/datasets/LSY-lab/libero_plus_goal_noise_all_sawyer_failures/tree/main/videos with the pattern videos/{video_key}/chunk-{chunk_index:03d}/file-{file_index:03d}.mp4. Available video keys are observation.images.frontview_image, observation.images.image, observation.images.sideview_image, and observation.images.wrist_image. Video frames align to numeric rows by episode/frame index at 20 FPS.

Reading the TsFile

Use the Apache TsFile SDK or a compatible table reader. For example, with the Python TsFile SDK:

from tsfile import TsFileReader
reader = TsFileReader('data/libero_plus_goal_noise_all_sawyer_failures.tsfile')
# Query/filter with the table reader API; episode_index and task_index are TAGs.

The mirrored meta/info.json records the source metadata and a tsfile_conversion object with the exact mapping, flattened features, row count, and original video policy.

Downloads last month
26