From Woozy Song@21:1/5 to All on Sat Oct 5 15:58:32 2024
Some useless information: Plots of test loss versus dataset size for
large language models give you an equation that computes the dimension
of natural language as 42.
Both OpenAI and Google research arrived at this.