What's Changed
- Add
serialization
andbeam
modules to the API. - Simplify github organization, so that all python modules are in
dataset_grouper/
. - Update script to build and publish package.
- Added utilities for creating test datasets.
- Added more examples, including a Dirichlet partitioning example.
- Removed
preprocess_fn
arg fromPartitionedDataset.build_group_stream
. Preprocessing should now be applied viatf.data.Dataset.map
.
Full Changelog: v0.2.1...v0.3.0