espnet2.train.gan_trainer.GANTrainerOptions

source

class espnet2.train.gan_trainer.GANTrainerOptions(ngpu: int, resume: bool, use_amp: bool, train_dtype: str, grad_noise: bool, accum_grad: int, grad_clip: float, grad_clip_type: float, log_interval: int | None, no_forward_run: bool, use_matplotlib: bool, use_tensorboard: bool, use_wandb: bool, adapter: str, use_adapter: bool, save_strategy: str, output_dir: Path | str, max_epoch: int, seed: int, sharded_ddp: bool, patience: int | None, keep_nbest_models: int | List[int], nbest_averaging_interval: int, early_stopping_criterion: Sequence[str], best_model_criterion: Sequence[Sequence[str]], val_scheduler_criterion: Sequence[str], unused_parameters: bool, wandb_model_log_interval: int, create_graph_in_tensorboard: bool, generator_first: bool, skip_discriminator_prob: float)

Bases: TrainerOptions

Trainer option dataclass for GANTrainer.

This class encapsulates the options that are specific to the GAN training process, inheriting from TrainerOptions. The options defined here include parameters that control the behavior of the generator and discriminator during training.

generator_first

Indicates whether to update the generator first during training.

• Type: bool

skip_discriminator_prob

The probability of skipping the discriminator update step. If this value is greater than 0, the discriminator will be updated with this probability.

• Type: float

Examples

To create an instance of GANTrainerOptions with specific settings:

python options = GANTrainerOptions(generator_first=True, skip_discriminator_prob=0.1)

This instance indicates that the generator should be updated first, and the discriminator update will be skipped with a probability of 0.1.

generator_first

*: bool*

skip_discriminator_prob

*: float*