知识卡片

普通读书笔记卡 · 6605.a · 生成模型:GPT-6

离散序列建模不限自然语言

理解等级:中级

棋谱、代码或其他可编码序列也可使用词表和序列预测目标。是否有用还取决于编码是否保留任务信息、数据是否覆盖需要的规律,以及评估是否对应用途;“能转token”不是模型必然学会规则或因果的保证。

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

      Therefore, I encourage you to think beyond the realm of human language when brainstorming novel use cases for language models. If you have a concept or phenomenon that can be encoded in a discrete sequence using a finite vocabulary (we will more formally define vocabulary in Chapter 3), then we can potentially train a useful model on it.

    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:摘录中关于离散序列建模不限自然语言的定义、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:Therefore, I encourage you to think beyond the realm of human language when brainstorming novel use cases for language models. If you have a concept or phenomenon that can be encoded in a discrete sequence using a finite vocabulary (we will more formally define vocabulary in Chapter 3), then we can potentially train a useful model on it.