Yeah, though I have not seen a formula which takes the number of expected inference runs into account for calculating the optimal data/parameter balance.