知识卡片

普通读书笔记卡 · 6606.h · 生成模型:GPT-6

角色指令与输入数据应分层

理解等级:高级

应用的高层行为约束、用户请求与工具返回承担不同职责。示例中两种位置表现相近,不证明所有模型都会忽略角色差异;应按实际接口和权限边界组织,并把来自文档或工具的文本视作需要解释的数据。

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

      high-level overarching behavior of an LLM, like “You are a financial expert well versed in writing formal reports.” If you are allowing your users to directly interact with the LLM, then the system prompt can be used to provide your own instruction to the LLM along with the user request. My experiments have shown that it doesn’t matter much if you place your instructions in

    参考来源

    - 位置:《Designing Large Language Model Applications》第01章《大语言模型入门》(来源项目:a19c7933-4eda-4176-aebd-30fe7e1ad292;源文件:extracted/Designing Large Language Model Applications/_epub-src/OEBPS/ch01.html) - 结论依据:摘录中关于角色指令与输入数据应分层的定义、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:high-level overarching behavior of an LLM, like “You are a financial expert well versed in writing formal reports.” If you are allowing your users to directly interact with the LLM, then the system prompt can be used to provide your own instruction to the LLM along with the user request. My experiments have shown that it doesn’t matter much if you place your instructions in