v3.2.1

What's Changed

• Bump torchmetrics from 0.5.1 to 0.11.0 by @dependabot in #30

New Contributors

• @dependabot made their first contribution in #30

Full Changelog: 3.2.0...3.2.1

v3.2.0

What's Changed

• Fix symmetry of computed covariance matrix by @borchero in #24
• Refactor CI and add pre-commit configuration by @borchero in #29
• Enable compatibility with torchmetrics >= 0.6.0 by @marcovarrone in #27

New Contributors

• @marcovarrone made their first contribution in #27

Full Changelog: 3.1.4...3.2.0

PyCave 3.1.4

What's Changed

• Fix k-means++ mini-batch initialization by @borchero in #22

Full Changelog: 3.1.3...3.1.4

PyCave 3.1.3

Features

• Add a covariances property to the GaussianMixtureModel

Dependencies

• Bump PyTorch Lightning to ^1.6.0
• Bump LightKit to ^0.4.4

PyCave 3.1.2

• Fixes #13

PyCave 3.1.1

• Bump lightkit dependency

PyCave 3.1.0

• Move core components to LightKit and integrate it

PyCave 3.0.1

• Generalize interface of TensorDataLoader to allow for multiple tensors
• Do not require a sampler to be passed to the initializer of TensorDataLoader

PyCave 3.0.0

This release is a complete rewrite of PyCave, building upon PyTorch Lightning.

• All models can be run across multiple GPUs and nodes, boosting performance even more
• Testing has been improved significantly, ensuring that results are at least as good as scikit-learn

PyCave 2.0.5

• Fix a bug for diagonal covariance matrices when the dimension is the same as the number of means