3 ms·
Are you familiar with what is called "the drunkards walk" Because if you think stochastic inputs will not unfortunately admit of less benign paths being taken i
by ggm 4y ago
Are you familiar with what is called "the drunkards walk" Because if you think stochastic inputs will not unfortunately admit of less benign paths being taken inside the dataset.. I think you're probably wrong.
I have very little doubt the primary problem in the GPT<x> model is going to remain: it is capable of reproducing highly believable crap. In a world of pizzagate, that has a risk of becoming highly weighted "I told you so" and self-reinforcing.