Help protect the Great Barrier Reef with TensorFlow on Kaggle Join Challenge

tff.simulation.MetricsManager

An abstract base class for metrics managers.

A tff.simulation.MetricsManager is a utility to save metric data across a number of rounds of some simulation.

Methods

clear_metrics

View source

Clear out metrics at or after a given starting round_num.

Note that since save_metrics is only compatible with nonnegative integer round numbers, clear_metrics(round_num=0) corresponds to clearing all metrics previously saved via save_metrics.

Args
round_num A nonnegative integer representing the starting round number for clearing metrics.

save_metrics

View source

Saves metrics data for a given round.

Note that different implementations may save metrics in markedly different ways, including skipping metrics depending on their type or round_num.

Args
metrics A mapping with string valued keys.
round_num A nonnegative integer representing the round number associated with metrics.