One day I will learn how LLMs work and will maybe find some utility for the smaller ones but dear god people need to shut up about ChatGPT. I never want to see that word again.
@cda How they work is just "autocomplete, but bigger". And the bigger makes them worse in pretty much every way, except deceiving people by making information-shaped garbage. See also: overfitting.
@cda what if someone will say that "LLM-which-cannot-be-named" failed miserably?)