`LightningODE`#

class scdiffeq.core.lightning_models._lightning_ode.LightningODE(latent_dim: int = 50, name: str | None = None, mu_hidden: ~typing.List[int] | int = [2000, 2000], mu_activation: str | ~typing.List[str] = 'LeakyReLU', mu_dropout: float | ~typing.List[float] = 0.2, mu_bias: bool = True, mu_output_bias: bool = True, mu_n_augment: int = 0, sde_type: str = 'ito', noise_type: str = 'general', backend: str = 'auto', train_lr: float = 0.0001, train_optimizer: ~torch.optim.optimizer.Optimizer = <class 'torch.optim.rmsprop.RMSprop'>, train_scheduler: ~torch.optim.lr_scheduler._LRScheduler = <class 'torch.optim.lr_scheduler.StepLR'>, train_step_size: int = 10, dt: float = 0.1, adjoint: bool = False, loading_existing: bool = False, *args, **kwargs)[source]#

Bases: BaseForwardMixIn, BaseLightningDiffEq

LightningODE

Extended description.

Parameters:

latent_dim (int, optional) – Description. Default is 50.
name (str, optional) – Description. Default is None.
mu_hidden (Union[List[int], int], optional) – Description. Default is [2000, 2000].
mu_activation (Union[str, List[str]], optional) – Description. Default is ‘LeakyReLU’.
mu_dropout (Union[float, List[float]], optional) – Description. Default is 0.2.
mu_bias (bool, optional) – Description. Default is True.
mu_output_bias (bool, optional) – Description. Default is True.
mu_n_augment (int, optional) – Description. Default is 0.
sde_type (str, optional) – Description. Default is ‘ito’.
noise_type (str, optional) – Description. Default is ‘general’.
backend (str, optional) – Description. Default is “auto”.
train_lr (float, optional) – Description. Default is 1e-4.
train_optimizer (Optimizer, optional) – Description. Default is torch.optim.RMSprop.
train_scheduler (torch.optim.lr_scheduler._LRScheduler, optional) – Description. Default is torch.optim.lr_scheduler.StepLR.
train_step_size (int, optional) – Description. Default is 10.
dt (float, optional) – Description. Default is 0.1.
adjoint (bool, optional) – Description. Default is False.
version (str, optional) – Description. Default is __version__.
loading_existing (bool, optional) – Whether to load an existing model. Default is False.

Return type:

None

__init__(latent_dim: int = 50, name: str | None = None, mu_hidden: ~typing.List[int] | int = [2000, 2000], mu_activation: str | ~typing.List[str] = 'LeakyReLU', mu_dropout: float | ~typing.List[float] = 0.2, mu_bias: bool = True, mu_output_bias: bool = True, mu_n_augment: int = 0, sde_type: str = 'ito', noise_type: str = 'general', backend: str = 'auto', train_lr: float = 0.0001, train_optimizer: ~torch.optim.optimizer.Optimizer = <class 'torch.optim.rmsprop.RMSprop'>, train_scheduler: ~torch.optim.lr_scheduler._LRScheduler = <class 'torch.optim.lr_scheduler.StepLR'>, train_step_size: int = 10, dt: float = 0.1, adjoint: bool = False, loading_existing: bool = False, *args, **kwargs) → None[source]#