← 返回刚才阅读的位置

知识卡片

普通读书笔记卡 · 8710.d · 生成模型:GPT-6

链式笔记区分答案与背景

理解等级:中级

CoN为每篇检索文档生成摘要并判断是否直接支持答案、只提供背景或不相关。对证据不足要保留未知,不能用主题支持替代具体立场;笔记自身也需核查,相关背景不构成缺失事实。

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

      want to rephrase it so that it is more amenable to being processed by the LLM. Another possible operation could be to filter out some of the retrieved texts based on certain rules. All of these are conducted during the refine stage. In this section, we will discuss two such techniques, summarization and chain-of-note. Let’s start with discussing how we can summarize the retrieved texts.

    参考来源

    - 位置:《Designing Large Language Model Applications》第12章《检索增强生成》(来源项目:a19c7933-4eda-4176-aebd-30fe7e1ad292;源文件:extracted/Designing Large Language Model Applications/_epub-src/OEBPS/ch12.html) - 结论依据:摘录中关于链式笔记区分答案与背景的定义、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:want to rephrase it so that it is more amenable to being processed by the LLM. Another possible operation could be to filter out some of the retrieved texts based on certain rules. All of these are conducted during the refine stage. In this section, we will discuss two such techniques, summarization and chain-of-note. Let’s start with discussing how we can summarize the retrieved texts.