Alexander Mathiasen comments on New Scaling Laws for Large Language Models