Alex Mallen
@alextmallen
Redwood Research (@redwood_ai)
Prev. @AiEleuther
가입 August 2021
344 팔로잉 중 1.1K 팬
@TheStalwart These are different concepts. Overfitting is an issue that happens with perfect training data. Reward hacking happens when training data is imperfect.
인기 있는 사용자
