Jan Leike：CoT 忠实度未随结果导向 RL 提升
Somewhat surprising that faithfulness of chain-of-thought doesn't improve much with outcome-based RL
AI 导读
有点意外，chain-of-thought 的忠实度并没有随着 outcome-based RL 提升多少
查看原帖New Anthropic research: Do reasoning models accurately verbalize their reasoning? Our new paper shows they don't. This casts doubt on whether monitoring chains-of-thought (CoT) will be enough to reliably catch safety issues. https://t.co/K3MrwqUXX9
正文
当前显示原文，中文全文译文尚未就绪。
Somewhat surprising that faithfulness of chain-of-thought doesn't improve much with outcome-based RL