Hacker News new | ask | show | jobs
by scotty79 721 days ago
The idea behind this research is to generate answer few times and if results are semantically vastly different from each other then probably they are wrong.