splisosm.logger

Training logger utilities for GLM/GLMM optimization.

Classes

PatienceLogger

Logger for tracking training patience and convergence.

Module Contents

class splisosm.logger.PatienceLogger(batch_size, patience, min_delta=1e-05, diagnose=False)

Logger for tracking training patience and convergence.

For training MultinomGLM and MultinomGLMM.

Parameters:

• batch_size (int) – Number of samples in the batch.

• patience (int) – Number of epochs to wait after the last significant improvement.

• min_delta (float) – Minimum change in the loss to qualify as an improvement.

• diagnose (bool) – Whether to store parameter changes during training.

get_params_iter()

Return stored parameters during training if diagnose is True.

Returns:

List of dictionaries containing loss and parameters for each sample, or None if diagnose is False.

Return type:

list[dict] or None

log(loss, params)

Log loss for a given epoch and update best parameters if improved.

Parameters:

• loss (Tensor) – Loss for the current epoch.

• params (dict[str, Tensor]) – Parameters for the current epoch.

Return type:

None

batch_size

best_epoch

best_loss

best_params = None

convergence

diagnose = False

epoch = 0

epochs_without_improvement

min_delta = 1e-05

patience