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.
EgoConseq-Bench data snapshot
Complete P_bench/data/ research snapshot dated 2026-09-11, paired with
the project code.
The benchmark tests spatial prediction under specified ego-motion from an initial
egocentric observation, body radius, camera configuration, and forward/turn actions.
The benchmark contains 4,999 Seen and 2,000 Unseen questions. Its eight tasks are:
| ID | Task | Seen | Unseen |
|---|---|---|---|
| A1 | Collision occurrence | 600 | 250 |
| A2 | First-collision action | 800 | 250 |
| A3 | First-contact object category | 599 | 250 |
| A4 | Intermediate target direction | 600 | 250 |
| B1 | Final target distance | 600 | 250 |
| B2 | Final target direction | 600 | 250 |
| B3 | Intermediate target distance | 600 | 250 |
| C1 | Future-view selection | 600 | 250 |
The full snapshot includes data/benchmark/, data/sft/seen_v2/,
data/metadata/, and data/analysis/, including saved model responses and
existing evaluation artifacts. These artifacts retain their recorded completion
status; this archive does not imply that every stored evaluation is complete.
External simulator/source-dataset installations and model weights are separate.
Source metadata can contain original absolute paths; configure external asset
locations through the code repository's .env.example when recollecting data.
Archive: 177,095,916,803 bytes (164.93 GiB) in 17 volumes. Verified 873,032 archive entries against the source directory, including 872,767 files and 2 symbolic links. The zstd integrity check and full tar traversal passed.
Restore
Install the Hugging Face CLI, GNU tar and zstd. This dataset is public; authentication is optional. Download every volume and the checksum file:
hf download syp115/EgoConseq-Bench --repo-type dataset \
--local-dir egoconseq-data-download
(cd egoconseq-data-download && sha256sum -c SHA256SUMS)
From a fresh code checkout's P_bench/ directory, concatenate the ordered volumes
and extract, adjusting the download directory below:
cat /path/to/egoconseq-data-download/data.tar.zst.part-* | zstd -d | tar -xf -
Extraction recreates data/ and overwrites matching paths. Volumes belong to one
archive and cannot be extracted separately. Internal symbolic links and hard links
are retained. See archive_manifest.json for exact sizes, entry counts, source
hashes, archive verification and the paired code commit. Read the inference and
evaluation READMEs in the code repository for environment setup and scoring rules.
- Downloads last month
- 324