Bug fixes in ModelTrainingManager

Full Changelog: v0.2.0...v0.2.1

Major updates in all modules (some compatbility breaking changes)

Full Changelog: v0.1.3...v0.2.0

Bug fixes and minor updates v0.1.3

This release features:

1. Some major bug fixes in the imports configuration
2. Refactoring of the loading function for torch models
3. Addition of functionalities in the conversion utils
4. A new prototype class for model explaination using Captum (WIP)
5. The first version of an argument parser for the optimization module
6. And finally, some reworking in ModelTrainingManager to load checkpoint states for training restart

Full Changelog: v0.1.1...v0.1.3

Bug fixes v0.1.2

• No compatibility breaking changes for all the stable modules.
• Implement CI/CD pipeline for build stage.
• Fix some bugs and unit tests.

First release v0.1.0

Initialize versioning of library to synchronize workflow with pypi. This release contains:

• Functionalities as per API documentation with some very basic examples (more to come).
• Some modules have been specifically developed and have associated unit tests, runnable with pytest.
• Many other functionalities have been developed and tested separately (e.g. for other specific applications) and integrated in PTAF. Unit tests may be currently missing for these.

Documentation is built using Sphinx and will be available hosted on ReadTheDocs. CI/CD foreseen for future releases.