Dataset Viewer
The dataset viewer is not available for this subset.
Cannot get the split names for the config 'default' of the dataset.
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.

PhysMani data for X-VLA

This dataset contains selected LeRobot data, metadata, and VelMap2D files from the physmani and physmani_velmap2d directories used for X-VLA experiments. Upload is ongoing until upload_manifest.json exists with complete: true. An incomplete upload is not a complete dataset.

Layout

  • physmani/: original metadata and archive files (large archives are split).
  • physmani_velmap2d/: frame mapping and training normalization statistics.
  • shards/physmani/xvla_lerobot*: converted LeRobot data.
  • shards/physmani/train and shards/physmani/val: legacy raw episode shards uploaded before the scope change; retained, but outside the completed dataset scope.
  • shards/physmani_velmap2d/train and shards/physmani_velmap2d/val: VelMap2D files.

Each part-*.tar under shards/ is an independent tar archive containing paths starting with physmani/ or physmani_velmap2d/. Extract all shards into the same destination directory to restore the original layout. Original files are preserved byte for byte. Source archive copies are excluded by the scope change below.

For example, from the downloaded repository directory:

mkdir -p restored
find shards -name 'part-*.tar' -print0 | while IFS= read -r -d '' shard; do
    tar -xf "$shard" -C restored
done

Copy the small metadata files from physmani/ and physmani_velmap2d/ into their corresponding restored directories. Files ending in .part-00000, .part-00001, etc. are ordered binary fragments of an existing source archive, not independent archives. Reassemble each such archive with cat in numeric filename order before extracting it. These archive copies are optional if the corresponding raw-data shards have already been restored.

Metadata is preserved as originally used, including machine-specific paths. Adjust those paths for your training machine. Older .bkup.json metadata is retained for provenance; use the non-backup metadata for current inputs. Runtime logs, upload-tool logs and caches are excluded.

The final upload manifest records the byte count and SHA-256 for each uploaded data file. Uploading this data does not establish model quality or equivalence between experiment variants.

Upload scope change (2026-10-02)

The owner stopped uploading raw physmani/train and physmani/val, and excluded 20260619_robot_data_v8_balanced_train.tar.gz and 20260619_robot_data_v8_balanced_val.tar.gz. Previously uploaded raw shards are retained; raw train is incomplete. complete: true in the final manifest refers only to the revised scope, not to a complete raw episode dataset. Converted LeRobot data, VelMap2D train/val, and metadata remain included. physmani/xvla_data.tar is also excluded; it finished uploading before the stop request was processed and was subsequently removed from the current repository revision.

Downloads last month
535