ndw’s avatarndw’s Twitter Archive—№ 33,192

  1. RT jessesheidlower: Shocking! Massive datasets use for ML are full of terrible labelling theregister.com/2021/04/01/mit_ai_accuracy/