Every 6.2 months doubles the required training compute for modern ML systemsforum.effectivealtruism.org2 points·ohlennart··0 commentsOpen articleSaveView on HN