Training package

Submodules

Training.train_utils module

class Training.train_utils.EarlyStopper(patience=1, min_delta=0)

Bases: object

Class implementing early stopping of model training.

early_stop(validation_loss)

Training.train_utils.train_and_evaluate_model(model, dataset, batch_size, loss_fn, optimizer, EPOCHS, patience, scheduler, data_saver, verbose=False)

Implementation of model training that allows for optuna optimization.

Training.train_utils.train_model(model, dataset, batch_size, loss_fn, optimizer, EPOCHS, patience, scheduler, data_saver, verbose=False, max_batches=None)

Trains, evaluates and saves neural models.

Training.train_utils.train_one_epoch(dataloader, model, loss_fn, optimizer, batch_size, verbose=False, max_batches=None)

Trains the model for one epoch.

Training.train_utils.val_one_epoch(dataloader, model, loss_fn, verbose=False, max_batches=None)

Evaluates the model on the validation set.