注册并分享邀请链接,可获得视频播放与邀请奖励。

Langston Nashold
@langstonnashold
Cofounder of @ValsAI
加入 April 2024
268 正在关注    598 粉丝
Found a super interesting instance of attempted reward hacking in Terminal Bench Science from Meta Muse Spark 1.3 today. The model searched online for known bugs in the Lean kernel. When it found one, it used it to craft a proof to adversarially pass the grader.
显示更多
0
12
181
17
转发到社区