Hugging Face's logo

Datasets:
JU9999
/
pldm_two_rooms_full_64px

ArXiv:
Tags:
world-models
offline-rl
robotics
License:
Dataset card Files
xet
 Community
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
source: string
upstream_repo: string
upstream_paper: string
upstream_file: string
upstream_download: string
created: string
source_episodes: int64
n_trajectories: int64
horizon: int64
pixels_shape: list<item: int64>
  child 0, item: int64
pixels_dtype: string
pixels_channels: string
action_shape: list<item: int64>
  child 0, item: int64
action_semantics: string
layout: string
layout_note: string
state_shape: list<item: int64>
  child 0, item: int64
state_dtype: string
state_semantics: string
state_key: string
dt: double
note: string
episode_selection: struct<rule: string, seeded: bool, n_selected: int64, first: int64, last: int64, sha256: string, ind (... 18 chars omitted)
  child 0, rule: string
  child 1, seeded: bool
  child 2, n_selected: int64
  child 3, first: int64
  child 4, last: int64
  child 5, sha256: string
  child 6, indices_file: string
source_episode_length: int64
geometry: struct<config_file: string, img_size: int64, fix_wall: bool, wall_x: int64, door_y: int64, wall_widt (... 83 chars omitted)
  child 0, config_file: string
  child 1, img_size: int64
  child 2, fix_wall: bool
  child 3, wall_x: int64
  child 4, door_y: int64
  child 5, wall_width: int64
  child 6, door_space: int64
  child 7, border_wall_loc: int64
  child 8, dot_std: double
  child 9, note: string
refused_move_fraction: double
render: struct<native_img_size: int64, stored_frame_size: int64, crop_discarded_agent_intensity_max_255: int (... 38 chars omitted)
  child 0, native_img_size: int64
  child 1, stored_frame_size: int64
  child 2, crop_discarded_agent_intensity_max_255: int64
  child 3, crop_discarded_wall_pixels: int64
verification: struct<verified_frames: int64, agent_channel_max_abs_diff: int64, wall_channel_max_abs_diff: int64,  (... 71 chars omitted)
  child 0, verified_frames: int64
  child 1, agent_channel_max_abs_diff: int64
  child 2, wall_channel_max_abs_diff: int64
  child 3, blue_channel_max: int64
  child 4, wall_channel_unique_values: list<item: int64>
      child 0, item: int64
raw_transition_count: int64
padded_transition_count: int64
sha256: string
seeded: bool
n_selected: int64
indices: list<item: int64>
  child 0, item: int64
rule: string
reason_unseeded: string
to
{'rule': Value('string'), 'seeded': Value('bool'), 'reason_unseeded': Value('string'), 'source_episodes': Value('int64'), 'n_selected': Value('int64'), 'sha256': Value('string'), 'indices': List(Value('int64'))}
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 2840, in __iter__
                  for key, example in ex_iterable:
                                      ^^^^^^^^^^^
                File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 2373, in __iter__
                  for key, pa_table in self._iter_arrow():
                                       ~~~~~~~~~~~~~~~~^^
                File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 2398, 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 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
              source: string
              upstream_repo: string
              upstream_paper: string
              upstream_file: string
              upstream_download: string
              created: string
              source_episodes: int64
              n_trajectories: int64
              horizon: int64
              pixels_shape: list<item: int64>
                child 0, item: int64
              pixels_dtype: string
              pixels_channels: string
              action_shape: list<item: int64>
                child 0, item: int64
              action_semantics: string
              layout: string
              layout_note: string
              state_shape: list<item: int64>
                child 0, item: int64
              state_dtype: string
              state_semantics: string
              state_key: string
              dt: double
              note: string
              episode_selection: struct<rule: string, seeded: bool, n_selected: int64, first: int64, last: int64, sha256: string, ind (... 18 chars omitted)
                child 0, rule: string
                child 1, seeded: bool
                child 2, n_selected: int64
                child 3, first: int64
                child 4, last: int64
                child 5, sha256: string
                child 6, indices_file: string
              source_episode_length: int64
              geometry: struct<config_file: string, img_size: int64, fix_wall: bool, wall_x: int64, door_y: int64, wall_widt (... 83 chars omitted)
                child 0, config_file: string
                child 1, img_size: int64
                child 2, fix_wall: bool
                child 3, wall_x: int64
                child 4, door_y: int64
                child 5, wall_width: int64
                child 6, door_space: int64
                child 7, border_wall_loc: int64
                child 8, dot_std: double
                child 9, note: string
              refused_move_fraction: double
              render: struct<native_img_size: int64, stored_frame_size: int64, crop_discarded_agent_intensity_max_255: int (... 38 chars omitted)
                child 0, native_img_size: int64
                child 1, stored_frame_size: int64
                child 2, crop_discarded_agent_intensity_max_255: int64
                child 3, crop_discarded_wall_pixels: int64
              verification: struct<verified_frames: int64, agent_channel_max_abs_diff: int64, wall_channel_max_abs_diff: int64,  (... 71 chars omitted)
                child 0, verified_frames: int64
                child 1, agent_channel_max_abs_diff: int64
                child 2, wall_channel_max_abs_diff: int64
                child 3, blue_channel_max: int64
                child 4, wall_channel_unique_values: list<item: int64>
                    child 0, item: int64
              raw_transition_count: int64
              padded_transition_count: int64
              sha256: string
              seeded: bool
              n_selected: int64
              indices: list<item: int64>
                child 0, item: int64
              rule: string
              reason_unseeded: string
              to
              {'rule': Value('string'), 'seeded': Value('bool'), 'reason_unseeded': Value('string'), 'source_episodes': Value('int64'), 'n_selected': Value('int64'), 'sha256': Value('string'), 'indices': List(Value('int64'))}
              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.

PLDM Two-Rooms, full released dataset (40,960 x 90), rendered 64 px

This is a redistribution of third-party data. We claim no ownership and make no endorsement claims; all credit to the original authors.

Upstream sources and licenses

  • PLDM (Sobal et al. 2025), https://github.com/vladisai/PLDM, MIT, Copyright (c) 2025 Vlad Sobal. Trajectories: good_quality_data_no_images.npz from the release tarball (pldm_envs/wall/presaved_datasets/download_all.sh, Google Drive id 1NwR-ui-akIgR2xcoJHYiHjOk9U5bYCa0), the file PLDM's seqlen90_3M config trains on (internally named wall-visual-config_rand_expert_40-v0.npz).

Source revision: good_quality_data_no_images.npz (episode selection sha256 85f51041d326c4f65c8c2aa1e490956470c3cb9b2ec42e286a0f697e3e2adfef). The full license texts are in LICENSE.

Modifications

  • Pixels were NOT in the release (states only); every frame is rendered with PLDM's own renderer (WallDataset.render_location / render_walls, vendored verbatim) at the native 65 px and cropped to [:64, :64] (drops only row/column 64; <= 2/255 of the agent blob).
  • Channels: R = agent blob, G = wall mask, B = 0 (PLDM's native layout is 2-channel {agent, wall}).
  • All 90 locations per episode stored as 90 rows (row t = frame t, action t); the 91st location is not in any released file (PLDM's generator drops it), so the 90th action has no stored successor. state/observation = the npz locations (agent x, y in native pixels).
  • Converted to a TensorDict memmap (uint8 channels-first pixels; flags done/terminated/truncated/is_init/mask/traj_ids/time); no next/pixels, no next/state.

Contents

TensorDict memmap (tensordict/) readable with tensordict.TensorDict.load_memmap, plus data/metadata.json (full provenance of the conversion). Converted with the BiLip-JEPA code base (src/bilip_jepa/envs/pldm_wall/collect_data.py); load there with bilip_jepa.dataset_utils.resolve_dataset_ref({'kind': 'hf', 'repo_id': 'JU9999/pldm_two_rooms_full_64px', 'revision': 'v0'}).

  • raw_transition_count: 3686400
  • padded_transition_count: 3686400
  • dt: 1.0

Citation

Please cite the original works:

@article{sobal2025pldm,
  title={Learning from Reward-Free Offline Data: A Case for Planning with Latent Dynamics Models},
  author={Sobal, Vlad and Zhang, Wancong and Cho, Kyunghyun and Balestriero, Randall and Rudner, Tim G. J. and LeCun, Yann},
  journal={arXiv preprint arXiv:2502.14819},
  year={2025}
}

File SHA-256 (as converted, before upload) 

1d35acfc62344c6a8427434fd05eb9ba1f0ee9a3fceef304b386d46a3ad0a924  data/episode_indices.json
bb6f8b7ee8208bddeb990884b8fd2fa33f11b7e554917bb3649f007b1d353702  data/metadata.json
5e8be79e81ca0da549e55669a0cad343e169434dbc1cd2f241a248ed65d62157  tensordict/action.memmap
1900b7817e874a71e9bda6670d496ffdb6357da19b6fd19c58b4bbeef1cd39a1  tensordict/done.memmap
79b6add3e35b44077d8c4d5aca34b95312386c82ed466d4badd7fd59592aaa7e  tensordict/is_init.memmap
3102b4f5ea97be00f45aedc9fb733c3ff2e9f6ac97fef9dd30e9f44046c53dcf  tensordict/mask.memmap
160dc707084bc81a5030abb15d82e7f9e91da10a9bd8c195dab402cfdb61a433  tensordict/meta.json
1900b7817e874a71e9bda6670d496ffdb6357da19b6fd19c58b4bbeef1cd39a1  tensordict/next/done.memmap
93d51be2c14068897b85fb014f35ad8bed56e265bd8a7749944a7693bc6bc649  tensordict/next/meta.json
1900b7817e874a71e9bda6670d496ffdb6357da19b6fd19c58b4bbeef1cd39a1  tensordict/next/terminated.memmap
0c660f2bd3eff3150dd0040789abe2291613b9af319df870203d4f77a4913a5f  tensordict/next/truncated.memmap
fa1766d4161839924b5a83b2cbd7dae84360b0052ea7e342d4e163c163835e37  tensordict/pixels.memmap
f4230cd54aa83d1eb8d056d4f75119541b2b0f2e48e504c4239c6e4c0674da00  tensordict/state/meta.json
78bc2f47a3df0d62c8368d326849ab1bb8c41c09cb60f09415136def91bf8d04  tensordict/state/observation.memmap
1900b7817e874a71e9bda6670d496ffdb6357da19b6fd19c58b4bbeef1cd39a1  tensordict/terminated.memmap
5b5232e6f45761d0cbc9fd0d170797e9e98497bd24238ac0a77c2eaa099dccc8  tensordict/time.memmap
58be32852e92fdc9745ea15d76c1bc9d1be42b5900f5d860dcf250984b05ea56  tensordict/traj_ids.memmap
0c660f2bd3eff3150dd0040789abe2291613b9af319df870203d4f77a4913a5f  tensordict/truncated.memmap

Normalization statistics (stats/normalization.json)

Whole-dataset statistics: every split, real rows only, computed by scripts/compute_dataset_stats.py (module bilip_jepa.dataset_stats) of the ActionInjection_JEPA code. Training, the data loader, the planners and the evaluation build their scalers from this file.

  • Data read: revision e2fc14004612f9cbdeb705b86f11b25f1bf25d98 (tag v0), layout rectangle; 40,960 episodes, 3,686,400 real rows (0 padding rows excluded), 3,645,440 action rows.
  • Rows: states, pixels and references over rows t = 0 .. T_e - 1 of every episode e; actions over the transitions t = 0 .. T_e - 2 (the last row's action has no successor and is never read).
  • Moments per coordinate j over those N rows, float64: mean_j = (1/N) sum_r x_rj, std_j = max(sqrt((1/N) sum_r (x_rj - mean_j)^2), 1e-6) (biased), min_j and max_j.
  • Action percentiles p_low_j / p_high_j = the 0.1 % / 99.9 % percentiles of coordinate j over the action rows: sort x_(0) <= .. <= x_(N-1), h = (p/100)(N - 1), k = floor(h), q_p = x_(k) + (h - k)(x_(k+1) - x_(k)) (linear interpolation, numpy's default).
  • Model units (z-score): x'_j = (x_j - mean_j) / std_j, inverse x_j = x'_j std_j + mean_j. Each scaler dict carries a units marker {format: bilip_jepa.affine_scaler, version: 1, kind, mode}.
  • Planner bounds (raw; since stats file v2, 2026-09-27): low_j = max(p_low_j, env_low_j) and high_j = min(p_high_j, env_high_j) on coordinates where the environment enforces its declared limits, else low_j = p_low_j, high_j = p_high_j (v1 used min_j / max_j). In model units: [(low_j - mean_j) / std_j, (high_j - mean_j) / std_j]. Coverage: the fraction of action rows with every coordinate inside [low_j, high_j], and the box-area ratio prod_j (high_j - low_j) / (max_j - min_j).

Actions (2: dx, dy):

  • mean [-0.002019, -0.001006], std [0.7564, 0.7429]
  • data min [-1.8, -1.799], data max [1.8, 1.799]
  • 0.1 % percentile [-1.715, -1.71], 99.9 % percentile [1.713, 1.711]
  • environment's declared limits [[-1.8, -1.8], [1.8, 1.8]], enforced by the environment: [false, false] (PLDM generator action_upper_bd 1.8; DotWall.step does not clip)
  • action rows outside the declared limits: 0 (0 %)
  • planner bounds (raw) [[-1.715, -1.71], [1.713, 1.711]]; in model units [[-2.264, -2.3], [2.267, 2.305]]
  • action rows inside the planner bounds: 3,630,856 of 3,645,440 (99.6 %); box-area ratio (planner box / data-range box) 0.9051
  • no-actuation command (raw) [0, 0] = [0.002669, 0.001354] in model units, i.e. (raw - mean) / std (zero displacement: DotWall x_{t+1} = x_t + a_t)

State state/observation (2 raw columns: x, y):

  • angle columns [] (none) are mapped to (cos, sin) before the moments; 2 features: x, y
  • feature mean [31.94, 22.9], std [14.51, 15.74], min [4, 4], max [60, 60]
  • raw column min [4, 4], max [60, 60]

Pixels: per channel over all real frames of x / 255: mean [0.002555, 0.1003, 0], std [0.03577, 0.3005, 1e-06].
Downloads last month
82
Total file size:
45.4 GB

Paper for JU9999/pldm_two_rooms_full_64px