tff.learning.templates.LearningProcessOutput

A structure containing the output of a LearningProcess.next computation.

state A structure that will be passed to invocation of LearningProcess.next. Not intended for inspection externally, contains implementation details of the process.
metrics Metrics derived from the applying the process to a given state and input. This output is intended for surfacing values to better understand the progress of a learning process, and is not directly intended to be used as input to LearningProcess.next. For example, this output may include classical metrics surrounding the quality of a machine learning model (eg. the average accuracy of a model across the participating clients), or operational metrics concerning systems-oriented information (eg. timing information and the amount of communication occurring between clients and server).