malt.trainer.get_default_trainer

malt.trainer.get_default_trainer(optimizer: str = 'Adam', learning_rate: float = 0.001, n_epochs: int = 10, batch_size: int = - 1, reduce_factor: float = 0.5, patience: int = 10, without_player: bool = False, min_learning_rate: float = 1e-06, no_validation_threshold=20)[source]

Get the default training scheme for models.

Parameters
  • optimizer (str) – Name of the optimizer. Must be an attribute of torch.optim

  • learning_rate (float) – Initial learning rate.

  • n_epochs (int) – Maximum epochs.

  • batch_size (int) – Batch size.

  • validation_split (float) – Proportion of validation set.

  • reduce_factor (float) – Rate of learning rate reduction.

Returns

Callable

Return type

Trainer function.