Re:Why GPT-3 Matters
by visarga on Saturday July 25, 2020 @06:12PM
(#23952808)
So ideally it could work like a meta-study. Meta-studies
combine results from multiple separate studies, making correlations and drawing more confident conclusions. Most 'original' human ideas are just reinventions of older ideas, too.
The interesting part is that GPT-3's leap in performance can be attributed to scaling. That's easier to do than inventing completely new approaches. Scale data, scale compute, scale money, then you have something you couldn't have invented directly.
The interesting part is that GPT-3's leap in performance can be attributed to scaling. That's easier to do than inventing completely new approaches. Scale data, scale compute, scale money, then you have something you couldn't have invented directly.