← 返回刚才阅读的位置

知识卡片

普通读书笔记卡 · 6653 · 生成模型:GPT-6

统计检索不再要求逐句写回应规则

理解等级:中级

给定输入与回应对,可寻找与用户输入相近的已有语句,再重用其回应。数据提供行为样例,减少逐条规则编写,但仅用精确匹配仍会被拼写变化和新表达限制;历史回应本身也需符合当前语境与质量要求。

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

      Imagine you have access to an enormous database of human dialog sessions—one that contains statements paired with responses from thousands, or even millions, of conversations. In this scenario, you could certainly build a chatbot by copying your user’s input and searching the database for the exact same string of characters. And then you could simply reuse one of the responses to that sentence or phrase

    参考来源

    - 位置:《Natural Language Processing in Action, Second Edition》第01章《会读写的机器与自然语言处理》(来源项目:a19c7933-4eda-4176-aebd-30fe7e1ad292;源文件:extracted/Natural Language Processing in Action, Second Edition/_epub-src/OEBPS/Text/chapter-1.html) - 结论依据:摘录中关于统计检索不再要求逐句写回应规则的定义、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:Imagine you have access to an enormous database of human dialog sessions—one that contains statements paired with responses from thousands, or even millions, of conversations. In this scenario, you could certainly build a chatbot by copying your user’s input and searching the database for the exact same string of characters. And then you could simply reuse one of the responses to that sentence or phrase