[stnmtn]

I don't know if I'm quite willing to go there yet, but LLMs are barrelling towards a level where, if the pace continues, they will easily be looked upon as the most import advancement of computing.

The fact that training a statistical model on terabytes of text can lead quite naturally and obviously to what is shown in the video in the webpage is shocking. The shocking part is how natural it feels, and how I know this is something that can work.

Congratulations to the teams that have been going down this path, this is wildly impressive stuff.