Trainer triggers

Interval

class chainer.training.triggers.IntervalTrigger(period, unit)

Trigger based on a fixed interval.

This trigger accepts iterations divided by a given interval. There are two ways to specify the interval: per iterations and epochs. Iteration means the number of updates, while epoch means the number of sweeps over the training dataset. Fractional values are allowed if the interval is a number of epochs; the trigger uses the iteration and epoch_detail attributes defined by the updater.

For the description of triggers, see get_trigger().

Parameters:

• period (int or float) – Length of the interval. Must be an integer if unit is 'iteration'.
• unit (str) – Unit of the length specified by period. It must be either 'iteration' or 'epoch'.

__call__(trainer)

Decides whether the extension should be called on this iteration.

Parameters:trainer (Trainer) – Trainer object that this trigger is associated with. The updater associated with this trainer is used to determine if the trigger should fire. Returns:

True if the corresponding extension should be invoked in this

iteration.

Return type:bool

ManualSchedule

class chainer.training.triggers.ManualScheduleTrigger(points, unit)

Trigger invoked at specified point(s) of iterations or epochs.

This trigger accepts iterations or epochs indicated by given point(s). There are two ways to specify the point(s): iteration and epoch. iteration means the number of updates, while epoch means the number of sweeps over the training dataset. Fractional values are allowed if the point is a number of epochs; the trigger uses the iteration and epoch_detail attributes defined by the updater.

Parameters:

• points (int, float, or list of int or float) – time of the trigger. Must be an integer or list of integer if unit is 'iteration'.
• unit (str) – Unit of the time specified by points. It must be either 'iteration' or 'epoch'.

__call__(trainer)

Decides whether the extension should be called on this iteration.

Parameters:trainer (Trainer) – Trainer object that this trigger is associated with. The updater associated with this trainer is used to determine if the trigger should fire. Returns:

True if the corresponding extension should be invoked in this

iteration.

Return type:bool

Minimum and maximum values

class chainer.training.triggers.MaxValueTrigger(key, trigger=(1, 'epoch'))

Trigger invoked when specific value becomes maximum.

For example you can use this trigger to take snapshot on the epoch the validation accuracy is maximum.

Parameters:

• key (str) – Key of value. The trigger fires when the value associated with this key becomes maximum.
• trigger – Trigger that decides the comparison interval between current best value and new value. This must be a tuple in the form of <int>, 'epoch' or <int>, 'iteration' which is passed to IntervalTrigger.

class chainer.training.triggers.MinValueTrigger(key, trigger=(1, 'epoch'))

Trigger invoked when specific value becomes minimum.

For example you can use this trigger to take snapshot on the epoch the validation loss is minimum.

Parameters:

• key (str) – Key of value. The trigger fires when the value associated with this key becomes minimum.
• trigger – Trigger that decides the comparison interval between current best value and new value. This must be a tuple in the form of <int>, 'epoch' or <int>, 'iteration' which is passed to IntervalTrigger.