Metric Tracker

class MetricTracker(metrics_to_track=None)

Bases: Callback

Lightning Callback for Metric Tracking.

Tracks specified metrics during training.

Parameters:

metrics_to_track (list[str], optional) – List of metrics to track. Defaults to train loss.

setup(trainer, pl_module, stage)

Called when fit, validate, test, predict, or tune begins.

Parameters:

• trainer (Trainer) – A Trainer instance.

• pl_module (SolverInterface) – A SolverInterface instance.

• stage (str) – Either ‘fit’, ‘test’ or ‘predict’.

on_train_epoch_end(trainer, pl_module)

Collect and track metrics at the end of each training epoch.

Parameters:

• trainer (pytorch_lightning.Trainer) – The trainer object managing the training process.

• pl_module – The model being trained (not used here).

property metrics

Aggregate collected metrics over all epochs.

Returns:

A dictionary containing aggregated metric values.

Return type:

dict