Repository navigation

All pull requests

Labels (39)Milestones (4)New pull request

Open

(619)

Closed

(4,672)

Fix async callable callbacks in map and filter
#8933GoldArowana opened Nov 2, 2026
Require dill>=0.3.6 on Python 3.11+
#8932behroozazarkhalili opened Nov 2, 2026Contributor
Fix bare filenames for indices caches
#8931Linxiushen opened Nov 2, 2026Contributor
Preserve _type feature columns in dataset card metadata
#8930linjiaye929 opened Nov 2, 2026
Fix non-batched Dataset.filter with arrow and polars formats
#8929astroknot-sheep opened Nov 2, 2026
Keep string values of mixed-type JSON fields instead of parsing them as JSON
#8928behroozazarkhalili opened Nov 2, 2026Contributor
Preserve nested fields when combining streaming datasets
#8927Aurantium-47 opened Nov 2, 2026
Raise in cast_column when the column is not in the dataset
#8925ANDROIDHASSAN opened Nov 1, 2026
Close decoder-owned streams when decoding remote PDFs
#8924Dodothereal opened Nov 1, 2026
Fix double nesting of list subfeatures in Sequence(dict)
#8923simpleqt opened Nov 1, 2026
1
Fix ClassLabel.int2str for one-shot iterables
#8922linjiaye929 opened Nov 1, 2026
Fix dead links to WebDataset, Spark and Albumentations docs
#8921erdogan-deniz opened Oct 31, 2026
Preserve concatenated table rows when pickling integer column selections
#8920ElMagAmine opened Oct 31, 2026
Fix map cache reuse across num_proc when the cache path has glob/regex special characters
#8918udsy19 opened Oct 31, 2026
Fix casting a nullable LargeList of ArrayXD to a fixed-length List
#8917udsy19 opened Oct 31, 2026
Fix default configuration selection when updating dataset cards
#8916DiadkoShmek opened Oct 31, 2026
fix: encode non-JSON strings per value in Json.cast_storage
#8915Steve-20-ai opened Oct 31, 2026
Fix empty typed Dataset save and reload
#8913alberttwong opened Oct 31, 2026
docs: link Redis ADBC Dataset integration guide
#8911alberttwong opened Oct 31, 2026
Docs: add missing imports in Use with Polars data extraction example
#8910punit857 opened Oct 30, 2026
Close xopen handles in xnumpy_load, xpandas_read_excel, xpyarrow_parquet_read_table, xsio_loadmat
#8908michaelkillgta opened Oct 29, 2026
Preserve on-the-fly transforms when flattening dataset indices
#8904Mixriaoliao opened Oct 29, 2026
Load Harbor multi-step tasks without a top-level instruction.md
#8903xlb3055 opened Oct 28, 2026
Allow reading prepared datasets from read-only shared caches
#8902Oscar-man-shrestha opened Oct 28, 2026
Use the same class names in every split in DatasetDict.class_encode_column
#8901mottopanikeiku opened Oct 28, 2026

Navigation Menu

Repository navigation

All pull requests

Fix async callable callbacks in map and filter

Require dill>=0.3.6 on Python 3.11+

Fix bare filenames for indices caches

Preserve `_type` feature columns in dataset card metadata

Fix non-batched Dataset.filter with arrow and polars formats

Keep string values of mixed-type JSON fields instead of parsing them as JSON

Preserve nested fields when combining streaming datasets

Raise in cast_column when the column is not in the dataset

Close decoder-owned streams when decoding remote PDFs

Fix double nesting of list subfeatures in Sequence(dict)

Fix ClassLabel.int2str for one-shot iterables

Fix dead links to WebDataset, Spark and Albumentations docs

Preserve concatenated table rows when pickling integer column selections

Fix map cache reuse across num_proc when the cache path has glob/regex special characters

Fix casting a nullable LargeList of ArrayXD to a fixed-length List

Fix default configuration selection when updating dataset cards

fix: encode non-JSON strings per value in Json.cast_storage

Fix empty typed Dataset save and reload

docs: link Redis ADBC Dataset integration guide

Docs: add missing imports in Use with Polars data extraction example

Close xopen handles in xnumpy_load, xpandas_read_excel, xpyarrow_parquet_read_table, xsio_loadmat

Preserve on-the-fly transforms when flattening dataset indices

Load Harbor multi-step tasks without a top-level instruction.md

Allow reading prepared datasets from read-only shared caches

Use the same class names in every split in DatasetDict.class_encode_column

Pull requests