The dataset viewer is not available for this subset.
Exception: SplitsNotFoundError
Message: The split names could not be parsed from the dataset config.
Traceback: 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/webdataset/webdataset.py", line 80, in _split_generators
raise ValueError(
...<2 lines>...
)
ValueError: The TAR archives of the dataset should be in WebDataset format, but the files in the archive don't share the same prefix or the same types.
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 68, 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.
geo-obs base data: training split
Rendered videos (SAPIEN) of one object moving by itself in a furnished room, often into another object, for fine-tuning image-to-video models on physically correct contact. Ten tasks (open a drawer / a door / a lid, push furniture into a gap, through a doorway, under a table, rotate furniture against a wall, tip an object over, push an object to a table edge, push an object behind another). Each record has a text prompt saying what moves and how far; whether it is blocked is decided by the scene geometry, and outcomes are balanced (e.g. BLOCK / REACH pairs).
This repository holds only the training split (base): 48,755 records, 1280x960, 72 frames at 24 FPS (3 s), left
camera, with one JSON per record (prompt, outcome, camera, per-frame poses).
Layout
data/<task folder>/base_NNN.tar each holds <task folder>/render/<name>.mp4 and <task folder>/meta/<name>.json
manifest/base.csv one row per record: task, folder, name, shard, mp4 / json bytes
manifest/shards.json per shard: records, bytes, sha256
env/flash_attn-2.8.3+cu126torch2.6-...whl the flash-attn wheel of the training environment (see the code package)
Unpack every shard into one folder:
for f in data/*/*.tar; do tar -xf "$f" -C /path/to/geo_obs_base_data; done
A record name reads <prefix>_base_<record id>__<scene>__<camera>__view-<easy|hard><score>_place-<easy|hard><score>__<OUTCOME>.
In each meta JSON, prompt is the text condition and outcome the result. The absolute paths inside the JSON files
point to the machine that rendered them and can be ignored.
| task | folder | records |
|---|---|---|
| open_drawer | open_drawer | 4,775 |
| open_hinge | open_hinge | 4,952 |
| through_door | through_door | 4,984 |
| under_table | under_table | 4,968 |
| arrange_furniture | arrange_furniture | 4,958 |
| lay_down | lay_down | 4,556 |
| rotate | rotate | 4,914 |
| hide_behind | hide_behind | 4,784 |
| push_to_edge | push_to_edge | 4,998 |
| inv_obs | invis_obs | 4,866 |
Terms
Research use only, non-commercial. The rendered objects come from PartNet-Mobility (SAPIEN), Artiverse (which collects models from sources such as Objaverse / Objaverse-XL, 3D-FUTURE and Amazon Berkeley Objects), YCB, ReplicaCAD and procedurally generated books. Several of these sources allow non-commercial research use only, and every object keeps its original license. This repository contains renders only, no 3D models.
- Downloads last month
- 99