Datasets:

Mark-ZJTang
/

InterACT-controlled-coverage8192

Error code:   StreamingRowsError
Exception:    CastError
Message:      Couldn't cast
schema_version: string
raw_manifest_schema_version: string
oracle_schema_version: string
generator: string
config: struct<groups: int64, seed: int64, frames: int64, substeps: int64, height: int64, width: int64, mujo (... 206 chars omitted)
  child 0, groups: int64
  child 1, seed: int64
  child 2, frames: int64
  child 3, substeps: int64
  child 4, height: int64
  child 5, width: int64
  child 6, mujoco_gl: string
  child 7, profile: string
  child 8, camera_protocol: string
  child 9, camera_generator_version: string
  child 10, groups_per_base_family: int64
  child 11, split_groups: struct<train: int64, val: int64, test: int64>
      child 0, train: int64
      child 1, val: int64
      child 2, test: int64
  child 12, state_protocol: string
manifest_sha256: string
oracle_report_sha256: string
next_stage: string
recipe: struct<camera_protocol: string, frames: int64, groups: int64, height: int64, mujoco_gl: string, prof (... 80 chars omitted)
  child 0, camera_protocol: string
  child 1, frames: int64
  child 2, groups: int64
  child 3, height: int64
  child 4, mujoco_gl: string
  child 5, profile: string
  child 6, seed: int64
  child 7, state_protocol: string
  child 8, substeps: int64
  child 9, width: int64
reuse_source: struct<manifest_sha256: string, path: string>
  child 0, manifest_sha256: string
  child 1, path: string
trajectory_rotation_degrees_max: double
scope: string
references: list<item: struct<groups: int64, path: string, sha256: string>>
  child 0, item: struct<groups: int64, path: string, sha256: string>
      child 0, groups: int64
      child 1, path: string
      child 2, sha256: string
trajectory_rotation_degrees_mean: double
initial_state_ranges: list<item: double>
  child 0, item: double
to
{'initial_state_ranges': List(Value('float64')), 'recipe': {'camera_protocol': Value('string'), 'frames': Value('int64'), 'groups': Value('int64'), 'height': Value('int64'), 'mujoco_gl': Value('string'), 'profile': Value('string'), 'seed': Value('int64'), 'state_protocol': Value('string'), 'substeps': Value('int64'), 'width': Value('int64')}, 'references': List({'groups': Value('int64'), 'path': Value('string'), 'sha256': Value('string')}), 'reuse_source': {'manifest_sha256': Value('string'), 'path': Value('string')}, 'scope': Value('string'), 'trajectory_rotation_degrees_max': Value('float64'), 'trajectory_rotation_degrees_mean': Value('float64')}
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 483, 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 2951, in __iter__
                  for key, example in ex_iterable:
                                      ^^^^^^^^^^^
                File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 2461, in __iter__
                  for key, pa_table in self._iter_arrow():
                                       ~~~~~~~~~~~~~~~~^^
                File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 2486, 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 547, in _iter_arrow
                  for key, pa_table in iterator:
                                       ^^^^^^^^
                File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 430, 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 2378, in table_cast
                  return cast_table_to_schema(table, schema)
                File "/usr/local/lib/python3.14/site-packages/datasets/table.py", line 2306, in cast_table_to_schema
                  raise CastError(
                  ...<3 lines>...
                  )
              datasets.table.CastError: Couldn't cast
              schema_version: string
              raw_manifest_schema_version: string
              oracle_schema_version: string
              generator: string
              config: struct<groups: int64, seed: int64, frames: int64, substeps: int64, height: int64, width: int64, mujo (... 206 chars omitted)
                child 0, groups: int64
                child 1, seed: int64
                child 2, frames: int64
                child 3, substeps: int64
                child 4, height: int64
                child 5, width: int64
                child 6, mujoco_gl: string
                child 7, profile: string
                child 8, camera_protocol: string
                child 9, camera_generator_version: string
                child 10, groups_per_base_family: int64
                child 11, split_groups: struct<train: int64, val: int64, test: int64>
                    child 0, train: int64
                    child 1, val: int64
                    child 2, test: int64
                child 12, state_protocol: string
              manifest_sha256: string
              oracle_report_sha256: string
              next_stage: string
              recipe: struct<camera_protocol: string, frames: int64, groups: int64, height: int64, mujoco_gl: string, prof (... 80 chars omitted)
                child 0, camera_protocol: string
                child 1, frames: int64
                child 2, groups: int64
                child 3, height: int64
                child 4, mujoco_gl: string
                child 5, profile: string
                child 6, seed: int64
                child 7, state_protocol: string
                child 8, substeps: int64
                child 9, width: int64
              reuse_source: struct<manifest_sha256: string, path: string>
                child 0, manifest_sha256: string
                child 1, path: string
              trajectory_rotation_degrees_max: double
              scope: string
              references: list<item: struct<groups: int64, path: string, sha256: string>>
                child 0, item: struct<groups: int64, path: string, sha256: string>
                    child 0, groups: int64
                    child 1, path: string
                    child 2, sha256: string
              trajectory_rotation_degrees_mean: double
              initial_state_ranges: list<item: double>
                child 0, item: double
              to
              {'initial_state_ranges': List(Value('float64')), 'recipe': {'camera_protocol': Value('string'), 'frames': Value('int64'), 'groups': Value('int64'), 'height': Value('int64'), 'mujoco_gl': Value('string'), 'profile': Value('string'), 'seed': Value('int64'), 'state_protocol': Value('string'), 'substeps': Value('int64'), 'width': Value('int64')}, 'references': List({'groups': Value('int64'), 'path': Value('string'), 'sha256': Value('string')}), 'reuse_source': {'manifest_sha256': Value('string'), 'path': Value('string')}, 'scope': Value('string'), 'trajectory_rotation_degrees_max': Value('float64'), 'trajectory_rotation_degrees_mean': 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.

InterACT controlled quartets: coverage8192

Complete archive. All 8192 payload hashes, NPZ CRCs and referenced array headers passed verification during transfer on 2026-10-12.

Historical 8192-group coverage_v1 synthetic camera × physics intervention dataset. Camera protocol: linear_v1, data profile: iid, seed: 2026090801.

8192 quartets / 1024 families / 32768 variants.
Train: 5120 quartets / 640 families; validation (val): 1536 / 192; test: 1536 / 192.
Every family has 8 quartets. Every quartet contains C0P0, C1P0, C0P1, C1P1.
Each RGB variant has 24 frames at 128 × 128 pixels, RGB uint8.
Actions, effects, initial states, contact labels, camera schedules and physical states are stored as named arrays in NPZ files. Labels provide supervision and are not tokenizer inputs.

Files and loading

Preserve groups/*.npz, raw_strict_manifest.jsonl, _SUCCESS, dataset_metadata.json, entry_report.json, oracle_report.json, and coverage_audit.json together. The trainer wrapper build_b3_strict_dataloader reads the raw manifest from strict_root. Array references specify NPZ paths, keys, shapes, dtypes, and payload SHA-256. Use numpy.load(path, allow_pickle=False).

from huggingface_hub import snapshot_download
strict_root = snapshot_download(
    repo_id="Mark-ZJTang/InterACT-controlled-coverage8192",
    repo_type="dataset",
)

Provenance

Original source prefix: tangzuojin/interveneact/strict_anchor/mujoco_8192_coverage_20260908_v1/. NAS mirror: interveneact_eval_artifacts/b12_scale8192_20260908_v1/data. Manifest SHA-256: 6592ec0067420036c40f521085c1f5d390c7abac1f12a5aee302fa88d6e5378a. Generator commit: 9b307a6ed79af17a54105a457171bd1675e38529.

The included coverage audit records reuse from the 4096-group version. Its archive/seed checks do not establish independence from all sources. Consult the original audit for exact scope. This dataset differs from the original B10 family1024 package.

Transfer verification checks payload and publication hashes, NPZ CRCs, array headers, splits, and family isolation. Original publication reports are retained without rerunning the simulation or independently reproducing its physics oracle. Original data files are preserved byte-for-byte; README and verification_report.json are added for this transfer. No license grant is inferred from this archival transfer.

Downloads last month: 32

Total file size:

6.73 GB