Early Stopping

class neuro_fuzzy_toolbox.training.early_stopping.EarlyStopping(patience, delta=0, last_state=False)

Bases: object

Early stopping mechanism for halting the training of a machine learning model (torch.nn.Module) when no sufficient improvement is observed.

__call__(model, loss, verbose=False)

Evaluates whether training should be stopped. If the stopping criterion is met, the stop attribute is updated to True.

Parameters:

• model (torch.nn.Module) – Model to evaluate.

• loss (float) – Current loss value of the model.

• verbose (bool) – If True, prints a warning message when early stopping is triggered. Defaults to False.

__init__(patience, delta=0, last_state=False)

Initializes a new EarlyStopping instance.

Parameters:

• patience (int) – Number of epochs without improvement before stopping training.

• delta (float) – Minimum improvement required to consider that the model has improved. Defaults to 0.

• last_state (bool) – If True, restores the last model state when stopping instead of the best state found during training. Defaults to False.

reset()

Resets the early stopping mechanism to its initial state.

property stop

Indicates whether training should be stopped.

Returns:

True if the stopping criterion has been met, False otherwise.

Return type:

bool