Hugging Face's logo

Datasets:
rexheng
/
semantic-space-artifacts

Languages:
English
Tags:
word-embeddings
glove
histwords
fasttext
License:
Dataset card Files
xet
 Community
Dataset Viewer
Duplicate
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:    CastError
Message:      Couldn't cast
n: int64
dim: int64
dtype: string
normalized: bool
cased: bool
skipped_malformed_lines: int64
n_shared_fit: int64
n_heldout: int64
R_orthogonal_atol_1e-3: bool
n_shared_total: int64
heldout_mean_cosine: double
to
{'n_shared_fit': Value('int64'), 'n_heldout': Value('int64'), 'heldout_mean_cosine': Value('float64'), 'n_shared_total': Value('int64'), 'R_orthogonal_atol_1e-3': Value('bool')}
because column names don't match
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/json/json.py", line 343, in _generate_tables
                  self._cast_table(pa_table, json_field_paths=json_field_paths),
                  ~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
                File "/usr/local/lib/python3.14/site-packages/datasets/packaged_modules/json/json.py", line 132, in _cast_table
                  pa_table = table_cast(pa_table, self.info.features.arrow_schema)
                File "/usr/local/lib/python3.14/site-packages/datasets/table.py", line 2378, in table_cast
                  return cast_table_to_schema(table, schema)
                File "/usr/local/lib/python3.14/site-packages/datasets/table.py", line 2306, in cast_table_to_schema
                  raise CastError(
                  ...<3 lines>...
                  )
              datasets.table.CastError: Couldn't cast
              n: int64
              dim: int64
              dtype: string
              normalized: bool
              cased: bool
              skipped_malformed_lines: int64
              n_shared_fit: int64
              n_heldout: int64
              R_orthogonal_atol_1e-3: bool
              n_shared_total: int64
              heldout_mean_cosine: double
              to
              {'n_shared_fit': Value('int64'), 'n_heldout': Value('int64'), 'heldout_mean_cosine': Value('float64'), 'n_shared_total': Value('int64'), 'R_orthogonal_atol_1e-3': Value('bool')}
              because column names don't match

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.

Semantic Space artifacts

Contents of data/processed/ for Semantic Space: memmap-ready embedding matrices, a compressed fastText OOV model, HistWords decade files, a 200k-row 3D layout, and published bias word lists.

This dataset is the runtime artifact store. A Space or local backend with HF_ARTIFACTS_REPO set downloads it on first boot when manifest.json is missing.

Licenses

Dual-licensed at the file level. The Hub card license is PDDL because GloVe and HistWords (the bulk of the bytes) are PDDL. Compressed fastText is CC BY-SA 3.0 and must stay attributed.

Path prefix License Source
glove840/ PDDL 1.0 Stanford GloVe 840B 300d
histwords/ PDDL 1.0 HistWords eng-all SGNS and COHA-word SGNS
fasttext/ft_cc.en.300_freqprune_400K_100K_pq_300.bin CC BY-SA 3.0 Facebook cc.en.300 via compress-fasttext
fasttext/ft_to_glove_R.f32.npy, fasttext/procrustes.json derived (fit on the two spaces above) Orthogonal Procrustes R
layout/ derived from GloVe 840B (PDDL) PCA-50 + UMAP-3D of the first 200,000 rows
wordlists/ research lists reproduced from published papers see citations
manifest.json PDDL checksums, shapes, licenses

Citations

File tree

File What it is
manifest.json Schema version, shapes, dtypes, licenses, SHA256 of every .npy / words.txt / fastText .bin.
glove840/vectors.f16.npy Full GloVe 840B 300d, float16, C-order, L2-normalised, mmapable. Shape (n, 300) with n > 2e6.
glove840/words.txt UTF-8, one token per line; line index = row index.
glove840/meta.json {n, dim, dtype, normalized, cased}.
fasttext/ft_cc.en.300_freqprune_400K_100K_pq_300.bin Product-quantised Common Crawl fastText (any-token OOV).
fasttext/ft_to_glove_R.f32.npy Orthogonal Procrustes matrix (300, 300) float32 mapping unit fastText → GloVe 840B.
fasttext/procrustes.json Fit size, held-out count, held-out mean cosine.
histwords/eng-all/{year}.f16.npy Decade SGNS, float16 L2-normalised; years 1800..1990 step 10.
histwords/eng-all/{year}.vocab.json JSON array of types, same order as that decade’s rows.
histwords/coha/{year}.f16.npy COHA-word SGNS; years 1830..2000 step 10.
histwords/coha/{year}.vocab.json Matching vocab arrays.
layout/pca50.joblib sklearn.decomposition.PCA(50) fit on GloVe rows [0, 200000).
layout/umap3d.joblib umap-learn UMAP-3D fit (absent if layout.method is pca3-fallback).
layout/umap3d.f32.npy Display coordinates (200000, 3) float32.
layout/layout_words.json The 200,000 display words (first 200k of words.txt).
layout/clusters.u8.npy k-means k=12 labels, uint8.
wordlists/weat.json WEAT batteries (Caliskan et al.).
wordlists/garg.json Garg et al. occupation / group lists.
wordlists/bolukbasi.json Bolukbasi gender pairs and occupations.
wordlists/charlesworth.json Charlesworth / PNAS group labels.
wordlists/default_axes.json Default SemAxis pole pairs for the UI.

Approximate total size: 3.1–3.3 GB.

Disclaimer

No word list is filtered. Bias metrics and historical slurs are corpus statistics from the cited papers, not endorsements.
Downloads last month
57
Total file size:
3.71 GB