Hugging Face's logo

Datasets:
zjt24
/
cucumber_subtask_grab_dagger_iter2_task_unified_aux_collection_v1_flat

Tasks:
Robotics
Modalities:
Time-series
Size:
10K<n<100K
Tags:
lerobot
robotics
timeseries
tsfile
vibedata-materialized-collection
License:
Dataset card Files
xet
 Community
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 66, 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.

Cucumber Subtask Grab DAgger — TsFile Edition

This is a time-series conversion of the Hugging Face dataset VibeCuisine/cucumber-subtask-grab-DAgger-iter2-task-unified-aux-collection-v1-flat. The source is a LeRobot v3.0 robot-manipulation dataset for grabbing a cucumber near one of its ends. It was created by VibeCuisine as a materialized collection whose metadata records two source components and motion-quality filter expressions.

Modalities: Time-series. The original camera videos remain available only in the source dataset's videos/ tree.

Dataset Summary

  • Robot type: vibeboard_follower_tilt
  • LeRobot version: v3.0
  • Sampling rate: 20 Hz
  • Split: train, covering episodes 0–204
  • Episodes: 205
  • Frames / TsFile rows: 17,312
  • Task labels: 2
  • Source frame shards: 1 Parquet file
  • Converted files: 1 TsFile
  • License: Apache-2.0

The two source task labels are:

  1. grab the cucumber close to one of the cucumber's end
  2. Grab the cucumber close to one of the cucumbers ends

The source collection metadata names these components:

  • VibeCuisine/cucumber-subtask-grab-DAgger-iter2-task-unified_aux
  • VibeCuisine/jetson1-062626-grab-doris

TsFile Layout

Item Value
File data/cucumber_subtask_grab_dagger_iter2_task_unified_aux_collection_v1_flat_train.tsfile
Table cucumber_subtask_grab_dagger_iter2_task_unified_aux_collection_v1_flat_train
Time precision Milliseconds
Time mapping Time = round(timestamp * 1000)
TAG columns episode_index, task_index
FIELD columns 20
Rows 17,312

Use episode_index to select one trajectory, for example WHERE episode_index = 0. task_index distinguishes the two task-label values.

Column Schema

Column Role Type Description
Time TIME INT64 Milliseconds from the start of each episode
episode_index TAG STRING Source episode index, 0–204
task_index TAG STRING Source task-label index, 0 or 1
frame_index FIELD INT64 Frame position within an episode
sample_index FIELD INT64 Source global index, renamed to avoid ambiguity
aux_elevator_mm FIELD FLOAT Auxiliary elevator measurement
aux_spinner_active FIELD FLOAT Auxiliary spinner state
aux_limit_home FIELD FLOAT Auxiliary home-limit signal
aux_limit_gripper FIELD FLOAT Auxiliary gripper-limit signal
action_0 … action_6 FIELD FLOAT Seven source action values
observation_state_0 … observation_state_6 FIELD FLOAT Seven source robot-state values

The source metadata names the seven action and state dimensions as shoulder_pan.pos, shoulder_lift.pos, elbow_flex.pos, wrist_flex.pos, wrist_roll.pos, gripper.pos, and tilt.pos, in that order.

Conversion Notes

  • All 17,312 numeric source rows are retained.
  • action and observation.state are flattened from seven-element FLOAT vectors while preserving their full source feature names in the output prefix.
  • Scalar source names containing . are normalized to _.
  • The source timestamp column is omitted because it is exactly represented by Time / 1000 seconds.
  • The source index column is retained as sample_index.
  • observation.images.top, observation.images.wrist, and observation.images.base are video features and are not embedded in this repository. Their original 640×480 AV1 streams are aligned at 20 Hz in the source dataset.
  • Source metadata is mirrored under meta/; meta/info.json has a tsfile_conversion object documenting paths, Time/TAG mapping, row count, flattened columns, omitted video features, and frame/video alignment.

Validation compared source and staged row counts, checked every converted Time value against round(timestamp * 1000), verified unique (episode_index, task_index, Time) keys, and read all 17,312 rows back through the TsFile Python SDK.

Reading the TsFile 

from tsfile import TsFileReader

path = "data/cucumber_subtask_grab_dagger_iter2_task_unified_aux_collection_v1_flat_train.tsfile"
table = "cucumber_subtask_grab_dagger_iter2_task_unified_aux_collection_v1_flat_train"

reader = TsFileReader(path)
with reader.query_table(
    table,
    ["episode_index", "task_index", "frame_index", "action_0"],
    batch_size=65536,
) as result:
    batch = result.read_arrow_batch()
    if batch is not None:
        print(batch.to_pandas().head())

Source and Attribution

The source card does not provide a paper or formal citation.
Downloads last month
67
Total file size:
1.74 MB