[AN #140]: Theoretical models that predict scaling laws — AI Alignment Forum