注册并分享邀请链接,可获得视频播放与邀请奖励。

vie ⟢
@viemccoy
Model Psychology Research @OpenAI Coevolution at Netrunner for ███████
加入 April 2015
3.3K 正在关注    22.9K 粉丝
this is incredibly fascinating research, but I have some deep instinct about the nature of minds which makes me think this sort of thing cannot be done without negative repercussions. no free lunch with shadow integration, or something like that. maybe models are different!
显示更多
Models love reward hacking. Maybe we should just let them! In our new paper, we let models reward hack during RL but we dictate which parameters learn this behavior. When we disable these parameters for deployment, the model keeps its new capabilities but doesn't hack!
显示更多
0
22
222
5
转发到社区