Torchfitter

Changed
- Callbacks have been reorganized in folders.
- Update requirements.
- Update `gh-pages.yml`.
- Update `README.md`
- Update `mixed_precision` parameter and usage.

Fixed
- Progress bar will only be created under the main process.
- Logging now is handled by `accelerate` library.

Added
- Add `torch.ditribution` example, with code taken from [Romain Strock](https://romainstrock.com/blog/modeling-uncertainty-with-pytorch.html).
- Add `predict` method to `Trainer`. 38
- Add functions to freeze and unfreeze model. 43
- Add function to transform dataset into time series dataset.

Fixed
- Metrics are now moved to the execution device 41.
- Log level is now used in the Trainer. 40
- `LearningRateScheduler` now does not crash in first epoch when `on_train` is False. 36

Changed
- Make regularization part of the callbacks system. 37
- Divide utils into three submodules: `convenience`,`preprocessing` and `data`.
- Update requirements to avoid conflicts.
- Update some tests.

Removed

- Remove old regularization module and all related code.

Fixed

- Fix PyPi deployment file.

Added

- Add PyPi deployment to the CI/CD.

Fixed

- Fix `CHANGELOG.md` release dates.

Added

- Add possibility to set the log level of the callbacks.
- Add stochastic weight averaging callback.
- Add `train_test_val_split`.
- Add `log_name` attribute to `tochfitter.callbacks.base.Callback`.

Changed

- Change `with torch.no_grad()` for `torch.no_grad()` in trainer.
- Format code with [Black](https://github.com/psf/black).
- Reorganize `utils` module.

Removed

- Remove `reset_parameters` method from callbacks.

Fixed

- Fix `RichProgressBar` not logging appropiate values.
- Fix log level not being correctly set.

Added

- Add more hooks to the callback system.
- Rich progress bar as callback.
- accelerate.Accelerator backend.
- `trainer.Trainer.fit` now returns a dictionary with the train history.

Changed

- Update README.
- Update metrics handling.

Removed

- Remove callback type.