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
task_id: string
episode_count: int64
control_count: int64
episodes: list<item: struct<id: string, steps: int64, source_terminal_reward: double, source_terminal_done: bo (... 186 chars omitted)
  child 0, item: struct<id: string, steps: int64, source_terminal_reward: double, source_terminal_done: bool, source_ (... 174 chars omitted)
      child 0, id: string
      child 1, steps: int64
      child 2, source_terminal_reward: double
      child 3, source_terminal_done: bool
      child 4, source_observation_timing: string
      child 5, joint_observation_same_state_max_error: double
      child 6, joint_observation_next_state_max_error: double
      child 7, model_sha256: string
      child 8, model_family_sha256: string
source: struct<task_id: string, repository: string, revision: string, path: string, url: string, bytes: int6 (... 18 chars omitted)
  child 0, task_id: string
  child 1, repository: string
  child 2, revision: string
  child 3, path: string
  child 4, url: string
  child 5, bytes: int64
  child 6, sha256: string
source_retrieval: struct<method: string, whole_source_sha256_recomputed: bool, expected_whole_source_sha256: string, s (... 131 chars omitted)
  child 0, method: string
  child 1, whole_source_sha256_recomputed: bool
  child 2, expected_whole_source_sha256: string
  child 3, source_etag: string
  child 4, received_bytes: int64
  child 5, block_size: int64
  child 6, ranges: list<item: struct<start: int64, end: int64, sha256: string>>
      child 0, item: struct<start: int64, end: int64, sha256: string>
          child 0, start: int64
          child 1, end: int64
          child 2, sha256: string
filename: string
bytes: int64
sha256: string
model_family_count: int64
referenced_image_bytes: int64
conversion_seconds: double
to
{'task_id': Value('string'), 'episode_count': Value('int64'), 'control_count': Value('int64'), 'episodes': List({'id': Value('string'), 'steps': Value('int64'), 'source_terminal_reward': Value('float64'), 'source_terminal_done': Value('bool'), 'source_observation_timing': Value('string'), 'joint_observation_same_state_max_error': Value('float64'), 'joint_observation_next_state_max_error': Value('float64'), 'model_sha256': Value('string'), 'model_family_sha256': Value('string')}), 'source': {'task_id': Value('string'), 'repository': Value('string'), 'revision': Value('string'), 'path': Value('string'), 'url': Value('string'), 'bytes': Value('int64'), 'sha256': Value('string')}, 'filename': Value('string'), 'bytes': Value('int64'), 'sha256': Value('string'), 'model_family_count': Value('int64'), 'referenced_image_bytes': Value('int64'), 'conversion_seconds': Value('float64')}
because column names don't match
Traceback:    Traceback (most recent call last):
                File "/src/services/worker/src/worker/utils.py", line 149, 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 129, 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 544, 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 3052, in __iter__
                  for key, example in ex_iterable:
                                      ^^^^^^^^^^^
                File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 2559, in __iter__
                  for key, pa_table in self._iter_arrow():
                                       ~~~~~~~~~~~~~~~~^^
                File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 2584, 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 558, in _iter_arrow
                  for key, pa_table in iterator:
                                       ^^^^^^^^
                File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 440, 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 361, 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 2381, in table_cast
                  return cast_table_to_schema(table, schema)
                File "/usr/local/lib/python3.14/site-packages/datasets/table.py", line 2309, in cast_table_to_schema
                  raise CastError(
                  ...<3 lines>...
                  )
              datasets.table.CastError: Couldn't cast
              task_id: string
              episode_count: int64
              control_count: int64
              episodes: list<item: struct<id: string, steps: int64, source_terminal_reward: double, source_terminal_done: bo (... 186 chars omitted)
                child 0, item: struct<id: string, steps: int64, source_terminal_reward: double, source_terminal_done: bool, source_ (... 174 chars omitted)
                    child 0, id: string
                    child 1, steps: int64
                    child 2, source_terminal_reward: double
                    child 3, source_terminal_done: bool
                    child 4, source_observation_timing: string
                    child 5, joint_observation_same_state_max_error: double
                    child 6, joint_observation_next_state_max_error: double
                    child 7, model_sha256: string
                    child 8, model_family_sha256: string
              source: struct<task_id: string, repository: string, revision: string, path: string, url: string, bytes: int6 (... 18 chars omitted)
                child 0, task_id: string
                child 1, repository: string
                child 2, revision: string
                child 3, path: string
                child 4, url: string
                child 5, bytes: int64
                child 6, sha256: string
              source_retrieval: struct<method: string, whole_source_sha256_recomputed: bool, expected_whole_source_sha256: string, s (... 131 chars omitted)
                child 0, method: string
                child 1, whole_source_sha256_recomputed: bool
                child 2, expected_whole_source_sha256: string
                child 3, source_etag: string
                child 4, received_bytes: int64
                child 5, block_size: int64
                child 6, ranges: list<item: struct<start: int64, end: int64, sha256: string>>
                    child 0, item: struct<start: int64, end: int64, sha256: string>
                        child 0, start: int64
                        child 1, end: int64
                        child 2, sha256: string
              filename: string
              bytes: int64
              sha256: string
              model_family_count: int64
              referenced_image_bytes: int64
              conversion_seconds: double
              to
              {'task_id': Value('string'), 'episode_count': Value('int64'), 'control_count': Value('int64'), 'episodes': List({'id': Value('string'), 'steps': Value('int64'), 'source_terminal_reward': Value('float64'), 'source_terminal_done': Value('bool'), 'source_observation_timing': Value('string'), 'joint_observation_same_state_max_error': Value('float64'), 'joint_observation_next_state_max_error': Value('float64'), 'model_sha256': Value('string'), 'model_family_sha256': Value('string')}), 'source': {'task_id': Value('string'), 'repository': Value('string'), 'revision': Value('string'), 'path': Value('string'), 'url': Value('string'), 'bytes': Value('int64'), 'sha256': Value('string')}, 'filename': Value('string'), 'bytes': Value('int64'), 'sha256': Value('string'), 'model_family_count': Value('int64'), 'referenced_image_bytes': Value('int64'), 'conversion_seconds': Value('float64')}
              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.

LIBERO-Genesis: unofficial resources and visual audit

Code and corrected-replay version 0.5.1a1. The original 130-task runtime bundle remains at v0.3.0a1; the 6,500-episode source conversion remains at v0.5.0a1. This is an independent research preview, not an endorsed or simulator-equivalent LIBERO release.

Code and installation · Visual audit · Notices

Corrections and limits

Visual review found clipped wrist-camera geometry and a spinning passive bottle in earlier successful clips. Version 0.5.1a1 restores source clipping planes, enables dedicated box-box detection in forward physics, makes original inferred inertias explicit, and restores the stove's state-dependent red display. The same-state camera comparison and isolated contact experiments are in the audit report.

The dedicated box detector is unsupported with gradients in Genesis 1.4.6. Differentiable contact remains experimental; these forward fixes are not a certificate of differentiable stability. Original source-action replay can still fail.

All 44 demonstration model families (1,210 body records, covering 130 tasks/6,500 source episodes) were checked against MuJoCo 2.3.7 inertias. This is a parameter audit, not 6,500 completed Genesis rollouts. We reran 260 episodes across ten selected tasks: 225 meet the final goal. Two of those have passive-object motion warnings, leaving a downloadable subset of 223 episodes from nine tasks, 29,759 controls. All 35 goal failures and flagged cases remain in the evidence.

Files

  • runtime-v0.3.0a1.tar.gz: immutable task definitions, named reset states, resource metadata and original-asset manifests. Original mesh/texture binaries are fetched from pinned upstream repositories, not mirrored here.
  • demonstrations/v0.5.0a1/: all 6,500 source demonstrations, 1,007,618 controls, about 0.93 GB. Original non-image arrays and per-episode models. Original RGB remains upstream and is available on demand (about 100 GB for all tasks). Original observation timing is post-action.
  • demonstrations/v0.5.1a1/verified/: corrected, screened Genesis transitions (nine HDF5 files, 38,399,827 bytes). Real robot observations/actions/next observations and qpos; no RGB. Models contain explicit original inertias.
  • demonstrations/v0.5.1a1/source_physics.json: original parser's inertial/clipping constants for source model families. The corrected loader applies these to the unchanged source conversion.
  • previews/v0.5.1a1/00_before_after.mp4: 16.8-second camera/bottle comparison.
  • previews/v0.5.1a1/00_all_22_inspection.mp4: 22 unfiltered review episodes, 204.85 seconds, dual views at 720p/20 fps; 16 goal successes and 6 failures.
  • previews/v0.5.1a1/libero_genesis_visual_inspection_v051.zip: all MP4s, comparison clips, index and provenance.
  • evidence-v0.5.1a1.tar.gz: final installed-wheel acceptance, inertial/visual audits, all corrected replay traces and diagnostic experiments.
  • Older versioned resources/evidence are retained. In particular, 0.5.0a1's goal-success subset and preview videos predate these repairs and include the reported visual/passive-motion defects.

libero-genesis demos prepare downloads the source conversion. libero-genesis demos prepare --verified selects the corrected subset in the new version's cache. --with-images fetches original MuJoCo RGB, not Genesis images. See the demonstration guide for source timing, formats and selection limits.

Provenance and credit

Original demonstrations: yifengzhu-hf/LIBERO-datasets, commit f13aa24a3da8c43c7225569f28c562979fa0e35a. Ten source HDF5 files were fully hash-verified; 120 were read through checked HTTP ranges. The expected whole-file SHA identifies those pinned sources but was not recomputed for range inputs. Every converted artifact has a complete checksum.

LIBERO code: 8f1084e3132a39270c3a13ebe37270a43ece2a01; robosuite 1.4.0: fbee5844ff5632f5b5698e204ec5357ca50be0df; Genesis runtime: 1.4.6. The consumer MJCF parser is pinned to MuJoCo 4.0.1, while original inertias were exported with 2.3.7.

Project code is MIT. Converted LIBERO data retains the original README's CC BY 4.0 dataset terms and attribution; the original Hugging Face card separately declares Apache-2.0 metadata, retained as provenance. Original asset terms are unchanged. Credit Liu et al., LIBERO: Benchmarking Knowledge Transfer for Lifelong Robot Learning (2023), and acknowledge Genesis and robosuite. See the notices for modifications and asset boundaries.

Downloads last month
105