5 ms·

the article doesn't seem to consider cost of hyperparameter optimization prior to the final training...