I wrote up a short post on large language model scaling laws. Let me know what you think!