HoLa: B-Rep Generation using a Holistic Latent Representation
Paper • 2504.14257 • Published
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/packaged_modules/json/json.py", line 291, in _generate_tables
pa_table = paj.read_json(
io.BytesIO(batch), read_options=paj.ReadOptions(block_size=block_size)
)
File "pyarrow/_json.pyx", line 364, in pyarrow._json.read_json
File "pyarrow/error.pxi", line 155, in pyarrow.lib.pyarrow_internal_check_status
return check_status(status)
File "pyarrow/error.pxi", line 92, in pyarrow.lib.check_status
raise convert_status(status)
pyarrow.lib.ArrowInvalid: JSON parse error: Column() changed from object to number in row 0
During handling of the above exception, another exception occurred:
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/json/json.py", line 101, in _split_generators
pa_table = next(iter(self._generate_tables(**splits[0].gen_kwargs, allow_full_read=False)))[1]
~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/usr/local/lib/python3.14/site-packages/datasets/packaged_modules/json/json.py", line 306, in _generate_tables
batch = json_encode_fields_in_json_lines(original_batch, json_field_paths)
File "/usr/local/lib/python3.14/site-packages/datasets/utils/json.py", line 113, in json_encode_fields_in_json_lines
examples = [json_encode_field(example, json_field_path) for example in examples]
~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/usr/local/lib/python3.14/site-packages/datasets/utils/json.py", line 34, in json_encode_field
return {**example, field: json_encode_field(example.get(field), json_field_path)}
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
TypeError: 'int' object is not a mapping
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.
HoLa-BRep: DeepCAD test set
This release contains the original 2,424-object DeepCAD test split used by the HoLa-BRep data interface, with historical geometry and multimodal condition assets. It is intended for evaluation and reproducibility of HoLa: B-Rep Generation using a Holistic Latent Representation.
Only the test split is included. There are no training or validation samples, checkpoints, or light-field-distance (LFD) features. The original test membership is preserved.
Download and extract
Install the Hugging Face CLI if needed:
pip install -U huggingface_hub
From your HoLa-BRep project directory:
hf download YuXingyao/HoLa-BRep \
data.zip manifest.json SHA256SUMS README.md \
--repo-type dataset --local-dir hola-test-download
(cd hola-test-download && sha256sum -c SHA256SUMS)
unzip hola-test-download/data.zip -d data
The ZIP contains organized_data/ and data_index/ directly. Extracting with -d data produces the paths expected by the HoLa-BRep repository:
data/
├── data_index/
│ └── deduplicated_deepcad_testing_7_30.txt
└── organized_data/
├── 00000093/
│ ├── data.npz
│ ├── img_feature_dinov2.npy
│ ├── imgs.npz
│ ├── normalized_shape.step
│ ├── pc.ply
│ ├── text.txt
│ └── text_feat.npy
└── ...
There is no data_lfd_feat/ directory in this release. Original LFD/novelty commands that need those features or training references are not supported by this package alone.
Files and layout
Each of the 2,424 IDs has seven files (16,968 object files in total). Their combined size before ZIP packaging is 10,533,747,354 bytes, approximately 10.53 GB / 9.81 GiB, excluding the split and release metadata.
| File | Contents |
|---|---|
data.npz |
Historical B-Rep arrays, including sample_points_faces, sample_points_lines, face_adj, edge_face_connectivity, and zero_positions. Objects have 7–30 faces. |
normalized_shape.step |
Historical normalized STEP geometry. |
imgs.npz |
RGB uint8 images at 224×224: 64 svr_imgs, 64 sketch_imgs, and 512 mvr_imgs. |
img_feature_dinov2.npy |
Float32 array (640, 1024): SVR rows 0–63, sketch rows 64–127, then eight MVR camera groups of 64 rows. |
pc.ply |
10,000 points with XYZ coordinates and normals. |
text.txt |
Text2CAD-derived descriptions associated with the CAD model. |
text_feat.npy |
Float32 array (4, 1024) containing precomputed text features. |
The 64-entry dimension follows the historical Euler augmentation enumeration, which includes repeated spatial rotations. It is not a set of 64 unique camera views. Within MVR, camera is the outer index and augmentation is the inner index.
manifest.json records every archive member's relative path, byte size and SHA256, plus the original ID order and archive information. SHA256SUMS covers the archive, manifest and this README.
Sources and licensing
This release preserves HoLa's original test membership from DeepCAD, together with historical HoLa preprocessing and condition assets. DeepCAD traces its CAD models to public Onshape documents linked by ABC. The historical text preparation pipeline uses annotations from Text2CAD. Please cite HoLa, DeepCAD and Text2CAD when using their assets.
Materials retain their applicable upstream terms; this packaging release does not relicense the underlying assets. The HoLa source repository's license file contains GNU GPL Version 3. Text2CAD's official dataset card specifies CC BY-NC-SA 4.0 for its annotations. Refer to the linked DeepCAD project for its data provenance and notices; its software license is not presented here as a separate license for all underlying CAD documents. The Hub metadata points to this section because the package contains materials from multiple upstream sources.
Release identity
- Release:
test-v1. - Test objects: 2,424; training/validation objects included: 0.
- Split file SHA256:
16352426a1878ee6836c3ddb306080a65313e92a5204a519e1386bdc59fc643a. - Historical loader reference: HoLa-BRep commit
8438eb328e01e3b54415e92370fac5dbec54a2df. - Paper: HoLa: B-Rep Generation using a Holistic Latent Representation.
- Downloads last month
- -