知识卡片

普通读书笔记卡 · 6498.b · 生成模型:GPT-6

强化学习以奖励更新策略

理解等级:中级

强化学习利用奖励引导模型提高某些输出或行动的概率,训练会改变权重。奖励可能来自规则、环境或验证器,定义什么被奖励直接决定优化方向;奖励提高不必然代表所有用户目标都改善。

这张卡片看不懂?

已有反馈 · 0 条

    还没有人提交反馈。

    在整书笔记中的位置

    原书出处 · 1 处

    • · 查看原书依据

      RL updates the model's weights during training. This enables the model to learn and refine reasoning strategies through trial and error, based on the feedback it receives from the environment. We will explore RL in more detail in chapters 6 and 7.

    参考来源

    - 位置:《Build a Reasoning Model (From Scratch)》第01章《理解推理模型》(来源项目:a19c7933-4eda-4176-aebd-30fe7e1ad292;源文件:extracted/Build a Reasoning Model (From Scratch)/_epub-src/OEBPS/Text/chapter-1.html) - 结论依据:摘录中关于强化学习以奖励更新策略的定义、比较或条件支撑卡片;原书表述不严谨的范围已在覆盖记录逐项限定,不把案例结果外推为普遍结论。 - 原始内容:RL updates the model's weights during training. This enables the model to learn and refine reasoning strategies through trial and error, based on the feedback it receives from the environment. We will explore RL in more detail in chapters 6 and 7.