All notable changes to this project will be documented in this file.
The format is based on Keep a Changelog, and this project adheres to Semantic Versioning (although, due to internal constraints, the initial public version is 3.0.0).
- Decorator framework for Gaussian Processes, including:
- Classification,
- Distributed GPs,
- Hierarchical GPs with Bayesian hyperparameters,
- Multitask GPs (EXPERIMENTAL),
- Variational inference,
- Compositional input warping for GPs,
- Input normalisation,
- Higher-rank features (EXPERIMENTAL).
- Optimisation features:
- Smart optimiser with early stopping,
- Learning rate optimisation,
- Learning rate scheduler integration.
- Synthetic datasets for testing.
- Worked notebook examples of major features.
- Detailed online documentation, complete with references.
- Near-complete unit test coverage.
- Comprehensive type hints, checked by beartype.
- Support for Python 3.9-3.12.
- This changelog, to make it easier for users and contributors to see precisely what notable changes have been made between each release of the project.
- And various contributor-facing amenities:
- Contributor guidelines,
- Automatic pre-commit checks, including formatting and linting,
- GH Actions to check unit tests and documentation build on each PR.
- Test coverage badge.