4.4.0
📦 datasetsView on GitHub →
✨ 3 features🐛 7 fixes🔧 6 symbols
Summary
This release introduces native support for loading NIfTI medical imaging datasets and allows explicit control over audio channel selection during casting. Several bug fixes address issues related to shuffling, array copying, and dependency compatibility.
✨ New Features
- Added support for loading NIfTI medical imaging datasets via load_dataset or Dataset.from_dict with the Nifti transformer.
- Added support for specifying the number of channels when casting an audio column using Audio(num_channels=N). Samples now have shape (num_channels, num_samples).
- Added support for Python 3.14.
🐛 Bug Fixes
- Fixed random seed handling during shuffle and interleave_datasets operations.
- Fixed CI issue related to compressionfs.
- Improved argument passthrough for `_batch_setitems()`.
- Fixed rendering of [!TIP] block in stream.shuffle documentation.
- Resolved ValueError: Unable to avoid copy while creating an array.
- Fixed issues when applying transforms to columns.
- Updated support to handle fsspec version 2025.10.0.
🔧 Affected Symbols
load_datasetDataset.from_dictNiftiAudio_batch_setitemsstream.shuffle