Issues

• IterableDataset.filter fails with arrow, pandas and polars formats

  huggingface/datasets#8724 · Filter by author felipebridge opened on Oct 2, 2026

• concatenate_datasets(axis=1) crashes on arrow-backed IterableDatasets with different lengths

  huggingface/datasets#8719 · Filter by author Zhuoxi2000 opened on Oct 1, 2026
  1

  

• Casting mixed JSON and plain strings to Json depends on row order and batch size

  huggingface/datasets#8714 · Filter by author Marcuswang0824 opened on Sep 30, 2026
  1

• JSON loader turns strings like "123" or "null" into other types in inferred Json columns

  huggingface/datasets#8713 · Filter by author u7k4rs6 opened on Sep 30, 2026

• string_to_dict maps fields to unrelated regex capture groups

  huggingface/datasets#8709 · Filter by author Afloat16 opened on Sep 29, 2026
  1

  

• IterableDataset skip/take fails when iterating a nonzero epoch

  huggingface/datasets#8706 · Filter by author galleonli opened on Sep 28, 2026

• Skip constructing disabled hf_tqdm progress bars in map_nested and _single_map_nested

  huggingface/datasets#8705 · Filter by author zhixiangli opened on Sep 28, 2026

• Avoid repeated np.lib.NumpyVersion(np.__version__) parsing per column/row in NumpyArrowExtractor._arrow_array_to_numpy

  huggingface/datasets#8704 · Filter by author zhixiangli opened on Sep 28, 2026

• Windows: load_dataset fails with WinError 206 for moderately deep cache_dir because the lock filename duplicates the whole path

  huggingface/datasets#8702 · Filter by author 825pranav opened on Sep 27, 2026

• Dataset.from_pandas and pandas-formatted map fail on ArrayXD columns: PandasArrayExtensionArray has no __arrow_array__

  huggingface/datasets#8699 · Filter by author JoeyTan21 opened on Sep 27, 2026

• Casting large_string / string_view to ClassLabel fails, so Dataset.from_pandas(df, features=ClassLabel) breaks with pandas 3

  huggingface/datasets#8681 · Filter by author lucaluo925 opened on Sep 24, 2026

• load_dataset("json") fails on mixed-shape JSONL when a chunksize boundary separates the shapes

  huggingface/datasets#8675 · Filter by author aksakalmustafa opened on Sep 23, 2026
  1

• push_to_hub() / to_parquet() leave nested list/struct columns uncompressed

  huggingface/datasets#8671 · Filter by author aksakalmustafa opened on Sep 23, 2026

• IterableDataset.shuffle() in 5.0 reports num_shards = min(files per source), leaving DataLoader workers and distributed nodes without data

  huggingface/datasets#8669 · Filter by author rwightman opened on Sep 22, 2026

• IterableDataset numpy format returns object instead of the declared dtype

  huggingface/datasets#8663 · Filter by author Yonghui-Lee opened on Sep 22, 2026

• TranslationVariableLanguages crashes on empty translations with declared languages

  huggingface/datasets#8659 · Filter by author HuaTNA opened on Sep 21, 2026

• Long lock filenames do not provide mutual exclusion across Python processes

  huggingface/datasets#8653 · Filter by author Yang1107-wzy opened on Sep 21, 2026

• FileLock wrapper drops timeout and every other constructor argument

  huggingface/datasets#8637 · Filter by author behroozazarkhalili opened on Sep 17, 2026
  2

• Probabilistic interleave without replacement samples empty datasets

  huggingface/datasets#8626 · Filter by author HaokaiDing opened on Sep 15, 2026

• DataFilesList.filter() leaves origin_metadata misaligned with the filtered files

  huggingface/datasets#8618 · Filter by author betacatsling opened on Sep 14, 2026

• SplitDict.copy() crashes when dataset_name is not set

  huggingface/datasets#8617 · Filter by author betacatsling opened on Sep 14, 2026
  1

• load_dataset on a dataset name with no namespace raises HfUriError

  huggingface/datasets#8612 · Filter by author ebarkhordar opened on Sep 12, 2026

• push_to_hub() keeps stale dataset_info.features when re-pushing a single split with different columns

  huggingface/datasets#8608 · Filter by author belambert opened on Sep 12, 2026

• load_dataset silently accepts out-of-range ClassLabel indices when the source dtype already matches

  huggingface/datasets#8596 · Filter by author caiotheodoro opened on Sep 10, 2026
  2

• IterableDataset.decode(num_threads) busy-polls its ThreadPool futures, starving the decode threads — 0 rows on remote audio; callback-based bridging restores the intended 20x

  huggingface/datasets#8595 · Filter by author HaukurPall opened on Sep 10, 2026