Skip to content

Losses

Source

Losses Pydantic configurations.

KLLossConfig

Bases: BaseModel

KL loss configuration.

aggregation = 'mean' class-attribute instance-attribute

Aggregation of the KL loss across different layers.

annealing = False class-attribute instance-attribute

Whether to apply KL loss annealing.

annealtime = 10 class-attribute instance-attribute

Number of epochs for which KL loss annealing is applied.

current_epoch = 0 class-attribute instance-attribute

Current epoch in the training loop.

free_bits_coeff = 0.0 class-attribute instance-attribute

Free bits coefficient for the KL loss.

loss_type = 'kl' class-attribute instance-attribute

Type of KL divergence used as KL loss.

rescaling = 'latent_dim' class-attribute instance-attribute

Rescaling of the KL loss.

start = -1 class-attribute instance-attribute

Epoch at which KL loss annealing starts.

LVAELossConfig

Bases: BaseModel

LVAE loss configuration.

denoisplit_weight = 0.9 class-attribute instance-attribute

Weight for the denoiSplit loss (used in the muSplit-deonoiSplit loss).

kl_params = KLLossConfig() class-attribute instance-attribute

KL loss configuration.

kl_weight = 1.0 class-attribute instance-attribute

Weight for the KL loss in the total net loss. (i.e., net_loss = reconstruction_weight * rec_loss + kl_weight * kl_loss).

loss_type instance-attribute

Type of loss to use for LVAE.

musplit_weight = 0.1 class-attribute instance-attribute

Weight for the muSplit loss (used in the muSplit-denoiSplit loss).

non_stochastic = False class-attribute instance-attribute

Whether to sample latents and compute KL.

reconstruction_weight = 1.0 class-attribute instance-attribute

Weight for the reconstruction loss in the total net loss (i.e., net_loss = reconstruction_weight * rec_loss + kl_weight * kl_loss).