Dataset Viewer
Duplicate
The dataset viewer is not available for this split.
Cannot load the dataset split (in streaming mode) to extract the first rows.
Error code:   StreamingRowsError
Exception:    CastError
Message:      Couldn't cast
filename: string
overview_description: string
category: string
content_name: string
source_metadata: string
dataset_name: string
source_repo: string
stats: struct<total_selected: int64, target: int64, by_category: struct<Other: int64, Object Manipulation:  (... 356 chars omitted)
  child 0, total_selected: int64
  child 1, target: int64
  child 2, by_category: struct<Other: int64, Object Manipulation: int64, Consuming: int64, Object Interaction: int64, Gestur (... 262 chars omitted)
      child 0, Other: int64
      child 1, Object Manipulation: int64
      child 2, Consuming: int64
      child 3, Object Interaction: int64
      child 4, Gestures: int64
      child 5, Basic Locomotion Neutral: int64
      child 6, Basic Locomotion Styles: int64
      child 7, Advanced Locomotion: int64
      child 8, Unusual Locomotion: int64
      child 9, Sports: int64
      child 10, Household: int64
      child 11, Communication: int64
      child 12, Baseline: int64
      child 13, Complex Actions: int64
      child 14, Stunts: int64
      child 15, Environments: int64
  child 3, demo_included: list<item: string>
      child 0, item: string
motions: list<item: struct<filename: string, move_name: string, content_name: string, category: string, packa (... 180 chars omitted)
  child 0, item: struct<filename: string, move_name: string, content_name: string, category: string, package: string, (... 168 chars omitted)
      child 0, filename: string
      child 1, move_name: string
      child 2, content_name: string
      child 3, category: string
      child 4, package: string
      child 5, archive_csv_path: string
      child 6, take_name: string
      child 7, take_actor: string
      child 8, take_date: string
      child 9, move_duration_frames: int64
      child 10, is_neutral: string
      child 11, is_mirror: string
      child 12, is_demo8: bool
selection: struct<strategy: string, category_quotas: struct<Object Manipulation: int64, Object Interaction: int (... 394 chars omitted)
  child 0, strategy: string
  child 1, category_quotas: struct<Object Manipulation: int64, Object Interaction: int64, Basic Locomotion Neutral: int64, Basic (... 262 chars omitted)
      child 0, Object Manipulation: int64
      child 1, Object Interaction: int64
      child 2, Basic Locomotion Neutral: int64
      child 3, Basic Locomotion Styles: int64
      child 4, Advanced Locomotion: int64
      child 5, Gestures: int64
      child 6, Unusual Locomotion: int64
      child 7, Sports: int64
      child 8, Household: int64
      child 9, Consuming: int64
      child 10, Communication: int64
      child 11, Baseline: int64
      child 12, Complex Actions: int64
      child 13, Stunts: int64
      child 14, Environments: int64
      child 15, Other: int64
  child 2, excluded_categories: list<item: string>
      child 0, item: string
  child 3, duration_band_source_frames: list<item: int64>
      child 0, item: int64
to
{'dataset_name': Value('string'), 'source_repo': Value('string'), 'source_metadata': Value('string'), 'selection': {'strategy': Value('string'), 'category_quotas': {'Object Manipulation': Value('int64'), 'Object Interaction': Value('int64'), 'Basic Locomotion Neutral': Value('int64'), 'Basic Locomotion Styles': Value('int64'), 'Advanced Locomotion': Value('int64'), 'Gestures': Value('int64'), 'Unusual Locomotion': Value('int64'), 'Sports': Value('int64'), 'Household': Value('int64'), 'Consuming': Value('int64'), 'Communication': Value('int64'), 'Baseline': Value('int64'), 'Complex Actions': Value('int64'), 'Stunts': Value('int64'), 'Environments': Value('int64'), 'Other': Value('int64')}, 'excluded_categories': List(Value('string')), 'duration_band_source_frames': List(Value('int64'))}, 'stats': {'total_selected': Value('int64'), 'target': Value('int64'), 'by_category': {'Other': Value('int64'), 'Object Manipulation': Value('int64'), 'Consuming': Value('int64'), 'Object Interaction': Value('int64'), 'Gestures': Value('int64'), 'Basic Locomotion Neutral': Value('int64'), 'Basic Locomotion Styles': Value('int64'), 'Advanced Locomotion': Value('int64'), 'Unusual Locomotion': Value('int64'), 'Sports': Value('int64'), 'Household': Value('int64'), 'Communication': Value('int64'), 'Baseline': Value('int64'), 'Complex Actions': Value('int64'), 'Stunts': Value('int64'), 'Environments': Value('int64')}, 'demo_included': List(Value('string'))}, 'motions': List({'filename': Value('string'), 'move_name': Value('string'), 'content_name': Value('string'), 'category': Value('string'), 'package': Value('string'), 'archive_csv_path': Value('string'), 'take_name': Value('string'), 'take_actor': Value('string'), 'take_date': Value('string'), 'move_duration_frames': Value('int64'), 'is_neutral': Value('string'), 'is_mirror': Value('string'), 'is_demo8': Value('bool')})}
because column names don't match
Traceback:    Traceback (most recent call last):
                File "/src/services/worker/src/worker/utils.py", line 147, in get_rows_or_raise
                  return get_rows(
                      dataset=dataset,
                  ...<4 lines>...
                      column_names=column_names,
                  )
                File "/src/libs/libcommon/src/libcommon/utils.py", line 272, in decorator
                  return func(*args, **kwargs)
                File "/src/services/worker/src/worker/utils.py", line 127, in get_rows
                  rows_plus_one = list(itertools.islice(safe_iter(ds, dataset=dataset), rows_max_number + 1))
                File "/src/services/worker/src/worker/utils.py", line 478, in safe_iter
                  yield from ds.decode(False) if ds.features else ds
                File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 2818, in __iter__
                  for key, example in ex_iterable:
                                      ^^^^^^^^^^^
                File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 2355, in __iter__
                  for key, pa_table in self._iter_arrow():
                                       ~~~~~~~~~~~~~~~~^^
                File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 2380, in _iter_arrow
                  for key, pa_table in self.ex_iterable._iter_arrow():
                                       ~~~~~~~~~~~~~~~~~~~~~~~~~~~~^^
                File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 536, in _iter_arrow
                  for key, pa_table in iterator:
                                       ^^^^^^^^
                File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 419, in _iter_arrow
                  for key, pa_table in self.generate_tables_fn(**gen_kwags):
                                       ~~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^
                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 2369, in table_cast
                  return cast_table_to_schema(table, schema)
                File "/usr/local/lib/python3.14/site-packages/datasets/table.py", line 2297, in cast_table_to_schema
                  raise CastError(
                  ...<3 lines>...
                  )
              datasets.table.CastError: Couldn't cast
              filename: string
              overview_description: string
              category: string
              content_name: string
              source_metadata: string
              dataset_name: string
              source_repo: string
              stats: struct<total_selected: int64, target: int64, by_category: struct<Other: int64, Object Manipulation:  (... 356 chars omitted)
                child 0, total_selected: int64
                child 1, target: int64
                child 2, by_category: struct<Other: int64, Object Manipulation: int64, Consuming: int64, Object Interaction: int64, Gestur (... 262 chars omitted)
                    child 0, Other: int64
                    child 1, Object Manipulation: int64
                    child 2, Consuming: int64
                    child 3, Object Interaction: int64
                    child 4, Gestures: int64
                    child 5, Basic Locomotion Neutral: int64
                    child 6, Basic Locomotion Styles: int64
                    child 7, Advanced Locomotion: int64
                    child 8, Unusual Locomotion: int64
                    child 9, Sports: int64
                    child 10, Household: int64
                    child 11, Communication: int64
                    child 12, Baseline: int64
                    child 13, Complex Actions: int64
                    child 14, Stunts: int64
                    child 15, Environments: int64
                child 3, demo_included: list<item: string>
                    child 0, item: string
              motions: list<item: struct<filename: string, move_name: string, content_name: string, category: string, packa (... 180 chars omitted)
                child 0, item: struct<filename: string, move_name: string, content_name: string, category: string, package: string, (... 168 chars omitted)
                    child 0, filename: string
                    child 1, move_name: string
                    child 2, content_name: string
                    child 3, category: string
                    child 4, package: string
                    child 5, archive_csv_path: string
                    child 6, take_name: string
                    child 7, take_actor: string
                    child 8, take_date: string
                    child 9, move_duration_frames: int64
                    child 10, is_neutral: string
                    child 11, is_mirror: string
                    child 12, is_demo8: bool
              selection: struct<strategy: string, category_quotas: struct<Object Manipulation: int64, Object Interaction: int (... 394 chars omitted)
                child 0, strategy: string
                child 1, category_quotas: struct<Object Manipulation: int64, Object Interaction: int64, Basic Locomotion Neutral: int64, Basic (... 262 chars omitted)
                    child 0, Object Manipulation: int64
                    child 1, Object Interaction: int64
                    child 2, Basic Locomotion Neutral: int64
                    child 3, Basic Locomotion Styles: int64
                    child 4, Advanced Locomotion: int64
                    child 5, Gestures: int64
                    child 6, Unusual Locomotion: int64
                    child 7, Sports: int64
                    child 8, Household: int64
                    child 9, Consuming: int64
                    child 10, Communication: int64
                    child 11, Baseline: int64
                    child 12, Complex Actions: int64
                    child 13, Stunts: int64
                    child 14, Environments: int64
                    child 15, Other: int64
                child 2, excluded_categories: list<item: string>
                    child 0, item: string
                child 3, duration_band_source_frames: list<item: int64>
                    child 0, item: int64
              to
              {'dataset_name': Value('string'), 'source_repo': Value('string'), 'source_metadata': Value('string'), 'selection': {'strategy': Value('string'), 'category_quotas': {'Object Manipulation': Value('int64'), 'Object Interaction': Value('int64'), 'Basic Locomotion Neutral': Value('int64'), 'Basic Locomotion Styles': Value('int64'), 'Advanced Locomotion': Value('int64'), 'Gestures': Value('int64'), 'Unusual Locomotion': Value('int64'), 'Sports': Value('int64'), 'Household': Value('int64'), 'Consuming': Value('int64'), 'Communication': Value('int64'), 'Baseline': Value('int64'), 'Complex Actions': Value('int64'), 'Stunts': Value('int64'), 'Environments': Value('int64'), 'Other': Value('int64')}, 'excluded_categories': List(Value('string')), 'duration_band_source_frames': List(Value('int64'))}, 'stats': {'total_selected': Value('int64'), 'target': Value('int64'), 'by_category': {'Other': Value('int64'), 'Object Manipulation': Value('int64'), 'Consuming': Value('int64'), 'Object Interaction': Value('int64'), 'Gestures': Value('int64'), 'Basic Locomotion Neutral': Value('int64'), 'Basic Locomotion Styles': Value('int64'), 'Advanced Locomotion': Value('int64'), 'Unusual Locomotion': Value('int64'), 'Sports': Value('int64'), 'Household': Value('int64'), 'Communication': Value('int64'), 'Baseline': Value('int64'), 'Complex Actions': Value('int64'), 'Stunts': Value('int64'), 'Environments': Value('int64')}, 'demo_included': List(Value('string'))}, 'motions': List({'filename': Value('string'), 'move_name': Value('string'), 'content_name': Value('string'), 'category': Value('string'), 'package': Value('string'), 'archive_csv_path': Value('string'), 'take_name': Value('string'), 'take_actor': Value('string'), 'take_date': Value('string'), 'move_duration_frames': Value('int64'), 'is_neutral': Value('string'), 'is_mirror': Value('string'), 'is_demo8': Value('bool')})}
              because column names don't match

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.

G1 BONES-SEED-100 (50 Hz)

A curated 100-motion Unitree G1 (29-DOF) reference set retargeted from the BONES-SEED motion-capture corpus, for imitation / tracking RL in IsaacLab-Imitation.

What's inside

npz/g1/*.npz                              100 G1 reference clips @ 50 Hz
manifests/g1_bones_seed_100_manifest.json training manifest (relative paths)
language/g1_bones_seed_100_language.json  natural-language descriptions + metadata
LANGUAGE_ANNOTATIONS.md                   language fields, ordering, and usage
curated/bones_seed_100_provenance.json    per-clip source provenance
curated/bones_seed_100_shortlist.timeline.json  selection shortlist

See LANGUAGE_ANNOTATIONS.md for the complete set of official BONES-SEED language options, their exact source columns, and the field used as the current planner command.

Selection

  • Diverse, locomotion + manipulation weighted: quotas biased toward Basic/Advanced Locomotion and Object Manipulation/Interaction, with coverage of Gestures, Sports, Household, Consuming, Communication, Stunts, and more.
  • Less dancing: the Dancing category is excluded.
  • Deduped: one canonical (non-mirror) clip per content_name.
  • Trainable in a propless env: aerial / ladder-climbing motions are excluded.
  • Subsumes the 8 demo motions used in prior closed-loop work (stoop, drinking, door handles, book reading, big-object manipulation, pick-up).

Reproduce the selection with scripts/select_bones_seed_100.py.

NPZ format

Each .npz contains (T = frames):

key shape frame
joint_pos, joint_vel (T, 29) joint space
qpos (T, 36) root(7) + joints(29)
qvel (T, 35) root(6) + joints(29)
root_pos, root_quat (T, 3), (T, 4) local frame
root_lin_vel, root_ang_vel (T, 3)
body_pos_w, body_quat_w, body_lin_vel_w, body_ang_vel_w (T, 32, ·) local frame
joint_names (29,) articulation order
body_names (32,) Isaac rigid-body order
fps (1,) 50

Joint order is the IsaacLab G1 articulation order (G1_29DOF_ISAACLAB_JOINT_NAMES, interleaved L/R breadth-first) and is embedded in every NPZ as joint_names — this is the order the sim consumes via write_joint_state_to_sim.

Local frame (no global / env_origin dependence): root_pos and body_pos_w share the native clip-local frame. Initial root XY remains within about 5 cm of the source origin. The env adds its own origin only while placing a reference in the scene; no Isaac scene-grid origin is stored in the NPZ.

Usage

pixi run -e isaaclab python scripts/rlopt/train.py \
    --task Isaac-Imitation-G1-Latent-v0 --algo IPMD --headless \
    env.lafan1_manifest_path=data/bones_seed_100/manifests/g1_bones_seed_100_manifest.json

Regenerate the private derived set from the locally available gated CSVs:

pixi run -e isaaclab python scripts/prepare_lafan1_from_csv.py \
    --csv_dir data/bones_seed/raw/g1_100 \
    --npz_dir data/bones_seed_100/npz/g1 \
    --manifest_path data/bones_seed_100/manifests/g1_bones_seed_100_manifest.json \
    --input_fps 120 --output_fps 50 --dataset_name bones_seed \
    --overwrite --headless --device cuda:0

pixi run python scripts/audit_bones_seed_phase5.py \
    --manifest data/bones_seed_100/manifests/g1_bones_seed_100_manifest.json \
    --report data/bones_seed_100/manifests/g1_bones_seed_100_preflight.json \
    --require-body-names

License / provenance

Derived from BONES-SEED (bones-studio/seed), which is gated and licensed — this derived G1 reference set inherits those terms and is kept private. Redistribution is subject to the upstream BONES-SEED license. Source clip identities are recorded in curated/bones_seed_100_provenance.json.

Downloads last month
94