scvi.train.SemiSupervisedTrainingPlanConfig

class scvi.train.SemiSupervisedTrainingPlanConfig(classification_ratio=50, lr=0.001, weight_decay=1e-06, n_steps_kl_warmup=None, n_epochs_kl_warmup=400, reduce_lr_on_plateau=False, lr_factor=0.6, lr_patience=30, lr_threshold=0.0, lr_scheduler_metric='elbo_validation', compile=False, compile_kwargs=None, loss_kwargs=<factory>)

Config for SemiSupervisedTrainingPlan.

Attributes table

classification_ratio
compile
compile_kwargs
lr
lr_factor
lr_patience
lr_scheduler_metric
lr_threshold
n_epochs_kl_warmup
n_steps_kl_warmup
reduce_lr_on_plateau
weight_decay
loss_kwargs

Methods table

to_kwargs()

Attributes

SemiSupervisedTrainingPlanConfig.classification_ratio: int = 50

SemiSupervisedTrainingPlanConfig.compile: bool = False

SemiSupervisedTrainingPlanConfig.compile_kwargs: dict | None = None

SemiSupervisedTrainingPlanConfig.lr: float = 0.001

SemiSupervisedTrainingPlanConfig.lr_factor: float = 0.6

SemiSupervisedTrainingPlanConfig.lr_patience: int = 30

SemiSupervisedTrainingPlanConfig.lr_scheduler_metric: Literal['elbo_validation', 'reconstruction_loss_validation', 'kl_local_validation'] = 'elbo_validation'

SemiSupervisedTrainingPlanConfig.lr_threshold: float = 0.0

SemiSupervisedTrainingPlanConfig.n_epochs_kl_warmup: int | None = 400

SemiSupervisedTrainingPlanConfig.n_steps_kl_warmup: int | None = None

SemiSupervisedTrainingPlanConfig.reduce_lr_on_plateau: bool = False

SemiSupervisedTrainingPlanConfig.weight_decay: float = 1e-06

SemiSupervisedTrainingPlanConfig.loss_kwargs: dict[str, Any]

Methods

SemiSupervisedTrainingPlanConfig.to_kwargs()

Return type:

dict[str, Any]