Papers like this shouldn’t be published openly. If such things significantly reduce the cost of training then it puts potentially dangerous technology within the hands of individuals anywhere. Rather, we need secure, regional GPU centers with international researchers publishing only to themselves providing beneficial services to the world and partner corporations.
Papers like this shouldn’t be published openly. If such things significantly reduce the cost of training then it puts potentially dangerous technology within the hands of individuals anywhere. Rather, we need secure, regional GPU centers with international researchers publishing only to themselves providing beneficial services to the world and partner corporations.