Understanding the differences between machine learning (ML) models is of
interest in scenarios ranging from choosing amongst a set of competing models,
to updating a deployed model with new training data. In these cases, we wish to
go beyond differences in overall metrics such as accur