avalanche.evaluation.metrics.Forgetting

class avalanche.evaluation.metrics.Forgetting[source]

The standalone Forgetting metric. This metric returns the forgetting relative to a specific key. Alternatively, this metric returns a dict in which each key is associated to the forgetting. Forgetting is computed as the difference between the first value recorded for a specific key and the last value recorded for that key. The value associated to a key can be update with the update method.

At initialization, this metric returns an empty dictionary.

__init__()[source]

Creates an instance of the standalone Forgetting metric

Methods

__init__()

Creates an instance of the standalone Forgetting metric

reset()

Resets the metric internal state.

reset_last()

result([k])

Forgetting is returned only for keys encountered twice.

update(k, v[, initial])

update_initial(k, v)

update_last(k, v)

Attributes

initial

The initial value for each key.

last

The last value detected for each key