Andrej Karpathy

@karpathy

Yeah, 95% of people misunderstand the tweet. I’m referring to gradient descent as a programmer (in the distributed representation space.) . In coding AI today the LLM is the programmer and in the regular “text space”. Ah well :)