← 返回刚才阅读的位置

知识卡片

普通读书笔记卡 · 8690.a · 生成模型:GPT-6

被动检索将证据选择交给检索器

理解等级:中级

系统根据查询检索文档或片段,装入上下文再要求模型回答。模型未主动选择检索步骤,但提示仍可包含来源信息,因此“被动”不意味着模型必然不知道来源;验收重点是检索充分性及回答忠实性。

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

      Passive Approach Figure 10-1 shows the typical workflow of an application that involves an LLM passively interacting with a data store. Figure 10-1. An LLM passively interacting with a data store A large number of use cases involve leveraging LLMs to use your own data. Examples include building a question-answering assistant over your company’s internal knowledge base that is spread over a bunch of Notion

    参考来源

    - 位置:《Designing Large Language Model Applications》第10章《外部工具与智能体》(来源项目:a19c7933-4eda-4176-aebd-30fe7e1ad292;源文件:extracted/Designing Large Language Model Applications/_epub-src/OEBPS/ch10.html) - 结论依据:摘录中关于被动检索将证据选择交给检索器的定义、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:Passive Approach Figure 10-1 shows the typical workflow of an application that involves an LLM passively interacting with a data store. Figure 10-1. An LLM passively interacting with a data store A large number of use cases involve leveraging LLMs to use your own data. Examples include building a question-answering assistant over your company’s internal knowledge base that is spread over a bunch of Notion