models/wavenet.py [193:231]:
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
    @staticmethod
    def default_config() -> ConfigProtocol:
        """
        Returns the OmegaConf config for this model.
        """
        return Config()

    def get_optimizers(
        self,
    ) -> List[
        Tuple[torch.optim.Optimizer, Optional[torch.optim.lr_scheduler._LRScheduler]]
    ]:
        """
        Get optimizers and schedulers used in this model. This is a method
        instead of just a pair of attributes because GANs tend to have
        multiple optimizers.

        Returns:
          A list of tuples. Each tuple contains an optimizer used by the
          model and an optional scheduler used for that optimizer. These are
          saved and restored from the checkpoints.
        """
        return [(self.optimizer, None)]

    def is_done(self) -> bool:
        """
        Checks if a model is done training.

        Returns:
          Whether the model is done training.
        """
        return self.global_step >= self.config.model.n_iterations

    def initialize(self) -> None:
        """
        Called after model creation.
        """

    def loss(self, spectrograms: Tensor, waveforms: Tensor) -> Tensor:
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -



models/wavernn.py [122:160]:
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
    @staticmethod
    def default_config() -> ConfigProtocol:
        """
        Returns the OmegaConf config for this model.
        """
        return Config()

    def get_optimizers(
        self,
    ) -> List[
        Tuple[torch.optim.Optimizer, Optional[torch.optim.lr_scheduler._LRScheduler]]
    ]:
        """
        Get optimizers and schedulers used in this model. This is a method
        instead of just a pair of attributes because GANs tend to have
        multiple optimizers.

        Returns:
          A list of tuples. Each tuple contains an optimizer used by the
          model and an optional scheduler used for that optimizer. These are
          saved and restored from the checkpoints.
        """
        return [(self.optimizer, None)]

    def is_done(self) -> bool:
        """
        Checks if a model is done training.

        Returns:
          Whether the model is done training.
        """
        return self.global_step >= self.config.model.n_iterations

    def initialize(self) -> None:
        """
        Called after model creation.
        """

    def loss(self, spectrograms: Tensor, waveforms: Tensor) -> Tensor:
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -