torchcell.losses.isomorphic_cell_loss.ICLoss

class ICLoss(lambda_dist: float, lambda_supcr: float, weights: Tensor | None = None)

Bases: Module

Weighted sum of MSE, distribution, and SupCR cell losses.

forward(predictions: Tensor, targets: Tensor, z_P: Tensor) → tuple[Tensor, dict[str, Any]]

Return the total weighted loss and a dict of component/normalized losses.