avalanche.evaluation.metrics.ExperienceBWT

class avalanche.evaluation.metrics.ExperienceBWT[source]

The Experience Backward Transfer metric.

This plugin metric, computed separately for each experience, is the difference between the last accuracy result obtained on a certain experience and the accuracy result obtained when first training on that experience.

This metric is computed during the eval phase only.

__init__()

Creates an instance of the ExperienceForgetting metric.

Methods

__init__()

Creates an instance of the ExperienceForgetting metric.

after_backward(strategy)

Called after criterion.backward() by the BaseStrategy.

after_eval(strategy)

Called after eval by the BaseStrategy.

after_eval_dataset_adaptation(strategy)

Called after eval_dataset_adaptation by the BaseStrategy.

after_eval_exp(strategy)

Called after eval_exp by the BaseStrategy.

after_eval_forward(strategy)

Called after model.forward() by the BaseStrategy.

after_eval_iteration(strategy)

Called after the end of an iteration by the BaseStrategy.

after_forward(strategy)

Called after model.forward() by the BaseStrategy.

after_train_dataset_adaptation(strategy)

Called after train_dataset_adapatation by the BaseStrategy.

after_training(strategy)

Called after train by the BaseStrategy.

after_training_epoch(strategy)

Called after train_epoch by the BaseStrategy.

after_training_exp(strategy)

Called after train_exp by the BaseStrategy.

after_training_iteration(strategy)

Called after the end of a training iteration by the BaseStrategy.

after_update(strategy)

Called after optimizer.update() by the BaseStrategy.

before_backward(strategy)

Called before criterion.backward() by the BaseStrategy.

before_eval(strategy)

Called before eval by the BaseStrategy.

before_eval_dataset_adaptation(strategy)

Called before eval_dataset_adaptation by the BaseStrategy.

before_eval_exp(strategy)

Called before eval_exp by the BaseStrategy.

before_eval_forward(strategy)

Called before model.forward() by the BaseStrategy.

before_eval_iteration(strategy)

Called before the start of a training iteration by the BaseStrategy.

before_forward(strategy)

Called before model.forward() by the BaseStrategy.

before_train_dataset_adaptation(strategy)

Called before train_dataset_adapatation by the BaseStrategy.

before_training(strategy)

Called before train by the BaseStrategy.

before_training_epoch(strategy)

Called before train_epoch by the BaseStrategy.

before_training_exp(strategy)

Called before train_exp by the BaseStrategy.

before_training_iteration(strategy)

Called before the start of a training iteration by the BaseStrategy.

before_update(strategy)

Called before optimizer.update() by the BaseStrategy.

metric_result(strategy)

metric_update(strategy)

reset()

Resets the metric.

reset_last()

Resets the last metric value.

result([k])

See Forgetting documentation for more detailed information.

update(k, v[, initial])

Update forgetting metric.

Attributes