how much real-world data do you think went into the evolution of the human brain and all its learning algorithms?
having 40 years of experience building software gives you no more insight into that than having 40 years of experience using language gives you insight into where your language skills come from
we're not sure if the actual data exactly matches our reconstruction, but one of the authors pointed out to us that we can exactly reproduce their scaling law if we make the mistake they made when fitting it to the data
what they did was to take the mean of the loss values across datapoints instead of summing them and used L-BFGS-B with the default tolerance settings, so the optimizer terminated early, and we can reproduce their results with this same mistake
one of their three approaches does not replicate and it's because of a software bug in the optimizer they used, i don't know what else we were supposed to say
how much real-world data do you think went into the evolution of the human brain and all its learning algorithms?
having 40 years of experience building software gives you no more insight into that than having 40 years of experience using language gives you insight into where your language skills come from