Hacker News new | ask | show | jobs
by pheggs 27 days ago
I dont buy that, mainly because in blind tests where people dont know which model was used and choose a result, this does not seem to hold true