Skip to content

changelog.md

Latest commit

 

History

History
47 lines (47 loc) · 1.61 KB

changelog.md

File metadata and controls

47 lines (47 loc) · 1.61 KB
Raw

Changelog

0.1.8

  • Fixed uploading large models to OpenML due to XML size limits in the flow
  • Added support for num_workers in dataloaders
  • Class weighting example
  • Fixed batch size bug
  • Fixed some small inconsistencies in the code
  • Fixed learner device str instead of torch.device (did not affect any actual results)
  • Removed some numpy and other bottlenecks for evaluation
  • Added automatic tests on python versions above 3.9

0.1.7, 3/04/25

  • Switched to poetry
  • Colab installation Fixed
  • Transformations can now also be applied separately on test
  • Much easier to change the optimizer and loss functions
  • Custom scheduler support
  • Argument support for loss, optim, schedulers
  • Hugging face models now work
  • Better progress bar
  • Much better documentation
  • Arbitrary layer support
  • Fixed old numpy imports
  • Added tests for specific types of tasks
  • Early Stopping example
  • BUG : Cannot upload large models to OpenML due to XML size limits in the flow

0.1.6, 21/03/25

  • Patch release to fix broken install on colab

0.1.5, 20/03/25

  • Patch release to fix broken install on colab

0.1.4, 20/03/25

  • Patch release to fix broken install on colab

0.1.3, 20/03/25

  • Add objects to openml runs

0.1.2, 13/03/25

  • Refactor for easier API use
  • Much cleaner documentation

0.1.1, 13/03/25

  • Added netron integration
  • Added tensorboard integration

0.1.0, 26/11/24

  • Complete overhaul of the trainer and data loader module
  • Better configuration
  • Refactor of examples
  • Fixed some callbacks
  • Created new datasets and tasks for image classification
  • Fixed Tabular datasets
  • Documentation using Github pages