Tweeted By @dennybritz
Poor graduate student: “Damn these labs with huge compute, I can’t compete with their model sizes! Well, at least with my small models I can focus on label efficiency! They generalize! Occam’s razor! Yay!”
— Denny Britz (@dennybritz) June 19, 2020
OpenAI and Google: “Bigger models are more label efficient”
😐