RT @davisblalock
"Language Models Can Teach Themselves to Program Better"

This paper changed my thinking about what future langauge models will be good at, mostly in a really concerning way. Let's start with some context: [1/11]