setu.NRETrainingResult

class setu.NRETrainingResult(nre, losses, val_losses, best_epoch)

Bases: object

Result of NRE training.

nre

Trained NRE instance with fitted transforms.

losses

Training loss per epoch.

val_losses

Validation loss per epoch.

best_epoch

Epoch with lowest validation loss.

Parameters:

• nre (NRE)

• losses (Array)

• val_losses (Array)

• best_epoch (int)

property estimator: NRE

The trained estimator (alias for .nre, uniform across result types).

validate(val_data, key, *, checks=None, strict=False, config=None)

Run validation checks on the trained NRE.

Parameters:

• val_data (ValidationDataset) – Validation dataset (separate from training data).

• key (Array) – JAX random key.

• checks (list[str] | None) – List of checks to run (default: all NRE checks).

• strict (bool) – If True, raise on failure.

• config (ValidationConfig | None) – Threshold configuration.

Return type:

ValidationResult