[GaggiX]

Well it's relevant if you want to compare the model trained optimally using the same amount of compute and this one parameter-bound to see how much you're trading.