lightning_template.utils.loop.kfold

Classes

KFoldLoop

Basic Loops interface.

Module Contents

class lightning_template.utils.loop.kfold.KFoldLoop(num_folds: int, fit_loop: lightning.pytorch.loops._FitLoop = None)

Bases: lightning.pytorch.loops._Loop

Basic Loops interface.

num_folds

current_fold = 0

_export_path = None

fit_loop = None

lightning_module_state_dict = None

property export_path: str

property done: bool

reset() → None

Nothing to reset in this loop.

on_run_start(*args: Any, **kwargs: Any) → None

Used to call setup_folds from the BaseKFoldDataModule instance and store the original weights of the model.

on_advance_start(*args: Any, **kwargs: Any) → None

Used to call setup_fold_index from the BaseKFoldDataModule instance.

advance(*args: Any, **kwargs: Any) → None

Used to the run a fitting and testing on the current hold.

on_advance_end() → None

Used to save the weights of the current fold and reset the LightningModule and its optimizers.

on_save_checkpoint() → Dict[str, int]

Called when saving a model checkpoint, use to persist loop state.

Returns:

The current loop state.

on_load_checkpoint(state_dict: Dict) → None

Called when loading a model checkpoint, use to reload loop state.

_reset_fitting() → None

__getattr__(key) → Any