datasets · Issues· 1386 open
Open on GitHubLocally synced open issues (discussions stay on GitHub)
- #8637
FileLock wrapper drops timeout and every other constructor argument
Updated Sep 17, 2026 - #887
pyarrow.lib.ArrowNotImplementedError: MakeBuilder: cannot construct builder for type extension<arrow.py_extension_type>
bugUpdated Sep 17, 2026 - #4981
Can't create a dataset with `float16` features
bugUpdated Sep 17, 2026 - #3638
AutoTokenizer hash value got change after datasets.map
bugUpdated Sep 17, 2026 - #6764
load_dataset can't work with symbolic links
enhancementUpdated Sep 17, 2026 - #5454
Save and resume the state of a DataLoader
enhancementgeneric discussionUpdated Sep 17, 2026 - #5481
Load a cached dataset as iterable
enhancementgood second issueUpdated Sep 17, 2026 - #5281
Support cloud storage in load_dataset
enhancementgood second issueUpdated Sep 17, 2026 - #8590
record_checksums default changed True -> False in 4.8.5, not in the release notes
Updated Sep 16, 2026 - #8596
load_dataset silently accepts out-of-range ClassLabel indices when the source dtype already matches
Updated Sep 16, 2026 - #5806
Return the name of the currently loaded file in the load_dataset function.
enhancementgood first issueUpdated Sep 15, 2026 - #8626
Probabilistic interleave without replacement samples empty datasets
Updated Sep 15, 2026 - #8381
IterableDataset.to_dict() and to_polars() return an empty generator instead of the data
Updated Sep 15, 2026 - #8618
`DataFilesList.filter()` leaves `origin_metadata` misaligned with the filtered files
Updated Sep 14, 2026 - #8617
`SplitDict.copy()` crashes when `dataset_name` is not set
Updated Sep 14, 2026