← 返回刚才阅读的位置

知识卡片

普通读书笔记卡 · 6788 · 生成模型:GPT-6

回复来源有不同的控制粒度

理解等级:中级

模板从已设计内容填值,检索从材料找候选,程序按规则计算,生成模型合成新文本。系统可混用这些路径,选择需看任务与可验证程度;检索排序或程序含随机性时,也不能统称全部确定性。

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

      Chatbots have exploded in popularity, as the tools for building them have started to generate uncanny simulations of intelligent human conversation. Several companies and platforms have been formed to help conversation designers build conversational assistants. The trendiest of these is the generative language models discussed in chapter 10. However, if you want to be able to maintain control over what your chatbot says, you will

    参考来源

    - 位置:《Natural Language Processing in Action, Second Edition》第12章《对话引擎与会话设计》(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-12.html) - 结论依据:摘录中关于回复来源有不同的控制粒度的定义、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:Chatbots have exploded in popularity, as the tools for building them have started to generate uncanny simulations of intelligent human conversation. Several companies and platforms have been formed to help conversation designers build conversational assistants. The trendiest of these is the generative language models discussed in chapter 10. However, if you want to be able to maintain control over what your chatbot says, you will