3 ms·
You know they are 1000% training these models to draw pelicans, this hasn't been a valid benchmark for 6 months +
by dakolli 6mo ago
You know they are 1000% training these models to draw pelicans, this hasn't been a valid benchmark for 6 months +
- simonw 6mo agoOpenAI must be very bad at training models to draw pelicans (and bicycles) then.
- Legend2440 6mo agoSkeptism is out of control these days, any time an LLM does something cool it must have been cheating.
- dakolli 6mo agothey legitimately suck at everything they don't have concrete examples to copy from.