Alex Mallen
@alextmallen
Redwood Research (@redwood_ai)
Prev. @AiEleuther
加入 August 2021
344 正在关注 1.1K 粉丝
@TheStalwart These are different concepts. Overfitting is an issue that happens with perfect training data. Reward hacking happens when training data is imperfect.
热门用户
