←新闻详情

@completeskeptic更新于 10月21日 22:46

Step-Jev:用通用验证器实现过程奖励

super sick! process rewards using a universal verifier (jev) - hopefully it's adversarially robust enough

AI 导读

太酷了!用通用验证器(jev)实现过程奖励——希望它的对抗鲁棒性足够强

引用

laxman

@llmluthor

Excited to release Step-Jev 🔥 step-by-step rewards for agent RL, even when there’s no answer key After DeepSeek R1, everyone said step rewards get hacked, so just reward the final answer. We found the opposite. With an AI judge grading only the final answer, our agent learned to fool it and accuracy dropped every single update until it was worse than no…

展开完整原文

Excited to release Step-Jev 🔥 step-by-step rewards for agent RL, even when there’s no answer key After DeepSeek R1, everyone said step rewards get hacked, so just reward the final answer. We found the opposite. With an AI judge grading only the final answer, our agent learned to fool it and accuracy dropped every single update until it was worse than no training at all. Step rewards fixed it. Everything is open!

查看原帖

正文

当前显示原文,中文全文译文尚未就绪。

super sick! process rewards using a universal verifier (jev) - hopefully it's adversarially robust enough

来源

@completeskeptic