Hacker News new | ask | show | jobs
by animal-husband 489 days ago
That is what was observed - o1 family models performed the “cheat”, non-reasoning models didn’t.