VIBE: Vector Index Benchmark for Embeddings
Paper • 2505.17810 • Published • 1
Dataset Viewer
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: ValueError
Message: Dataset 'gt_image' has length 1000 but expected 122287
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/hdf5/hdf5.py", line 76, in _generate_tables
num_rows = _check_dataset_lengths(h5, self.info.features)
File "/usr/local/lib/python3.14/site-packages/datasets/packaged_modules/hdf5/hdf5.py", line 353, in _check_dataset_lengths
raise ValueError(f"Dataset '{path}' has length {dset.shape[0]} but expected {num_rows}")
ValueError: Dataset 'gt_image' has length 1000 but expected 122287Need help to make the dataset viewer work? Make sure to review how to configure the dataset viewer, and open a discussion for direct support.
sc4020
Prepared benchmark data for SC4020 Project 1 (similarity search with text→image queries). Code: https://github.com/pbcong/sc4020
hf download pbcong/sc4020 --repo-type dataset --local-dir $SC4020_DATA
| File | Corpus | Queries | Vectors |
|---|---|---|---|
coco/coco_prepared.hdf5 |
122,287 COCO 2014 images | 1000 images + their first captions | CLIP ViT-B/32 (OpenAI), 512-d |
laion/laion_prepared.hdf5 |
999,448 LAION images | 1000 held-out images + 1000 VIBE text queries | CLIP ViT-B/32, 512-d |
Each HDF5 file holds corpus, queries_image, queries_text, gt_image, gt_text (exact top-100 ids),
gt_*_dist, and learn_text. Vectors are unit norm; similarity is inner product.
coco/coco_meta.json maps rows to COCO file names and query captions.
Sources and licences
- LAION subset from the VIBE benchmark (Jääsaari et al., 2025, arXiv:2505.17810), derived from LAION-400M embeddings, CC BY 4.0.
- MS-COCO 2014 (Lin et al., 2014): captions CC BY 4.0; embeddings computed by us with open_clip
ViT-B-32-quickgelu/openai. No images are included.
- Downloads last month
- 15