Jake Ward(@_jake_ward):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!

Jake Ward

@_jake_ward

i'm trying to figure out the computer

가입 February 2019

363 팔로잉 중 1.2K 팬

Jake Ward@_jake_ward

2026.10.19 19:26

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!