pinnx.callbacks module

This module provides callback functions for monitoring and controlling the training process. Callbacks can be used for model checkpointing, early stopping, logging, visualization, and adaptive training strategies.

Callbacks

Classes for training monitoring and control.

Callback: Base class for all callbacks.

CallbackList: Container for managing multiple callbacks.

ModelCheckpoint: Saves model checkpoints during training.

EarlyStopping: Stops training when monitored metric stops improving.

Timer: Tracks and reports training time.

DropoutUncertainty: Estimates prediction uncertainty using dropout.

VariableValue: Monitors and logs variable values during training.

OperatorPredictor: Makes predictions with trained operator networks.

MovieDumper: Creates animations of solution evolution.

PDEPointResampler: Adaptively resamples collocation points based on residuals.

Callback	Callback base class.
CallbackList	Container abstracting a list of callbacks.
ModelCheckpoint	Save the trainer after every epoch.
EarlyStopping	Stop training when a monitored quantity (training or testing loss) has stopped improving.
Timer	Stop training when training time reaches the threshold.
DropoutUncertainty	Uncertainty estimation via MC dropout.
VariableValue	Get the variable values.
OperatorPredictor	Generates operator values for the input samples.
MovieDumper	Dump a movie to show the training progress of the function along a line.
PDEPointResampler	Resample the training points for PDE and/or BC losses every given period.