Pipeline: - syncing with cloud - uploading split datasets for consistency between local and hosted
API: - version checking. Upgrading package version will be required for certain API changes. - improved error logging
Layers: - Lattice layer with hypercube interpolation (no monotonicity or simplex yet).
0.4.3
Bug Fix: - Restrict Pydantic version <2.0.0 due to use of `model_config` attribute in `TrainedModelMetadata` type.
0.4.2
Bug Fix: - Update use of `const=` in Pydantic `Field` to instead use `Literal` to fix import error. Requires updating the Pydantic version to >=1.10.8
0.4.1
0.4.0
Features: - Training now prints a progress bar so you no longer have to wait blindly. - `demo.py` containing a function for loading the UCI Statlog (Heart) demo data cleaned and ready.
0.3.2
Bug Fixes: - fix typo CATEGORICAL -> categorical that was breaking loading from hosted.