callcut.training.MetricsHistoryCallback

class callcut.training.MetricsHistoryCallback

Callback that records training metrics history.

Stores metrics at the end of each epoch for later analysis or plotting.

Attributes

history

Dictionary mapping metric names to lists of values per epoch.

Methods

on_train_epoch_end(trainer, pl_module)

Record metrics at the end of each training epoch.

Examples

>>> from callcut.training import MetricsHistoryCallback
>>> import lightning as L
>>>
>>> history_callback = MetricsHistoryCallback()
>>> trainer = L.Trainer(
...     max_epochs=10,
...     callbacks=[history_callback],
... )
>>> trainer.fit(module, datamodule=dm)
>>> history_callback.history["val_f1"]
[0.72, 0.78, 0.81, ...]

on_train_epoch_end(trainer, pl_module)

Record metrics at the end of each training epoch.

Parameters:

trainerTrainer

The Lightning trainer instance.

pl_moduleLightningModule

The Lightning module being trained.

property history

Dictionary mapping metric names to lists of values per epoch.

Type:

dict