知识卡片
因果脚手架分解文本机制
将King、Prince、Kingdom建成K→P、K→Y、P→Y的图,再分别学习p(K)、p(P|K)、p(Y|K,P),可把长故事生成分解为模块。图给出模型内的机制替换语义,但其真实因果含义仍取决于数据生成假设和模块是否充分。
这张卡片看不懂?
这张卡片已通过人工审查,不再接受失败反馈。
已有反馈 · 0 条
还没有人提交反馈。
在整书笔记中的位置
- · 因果与大语言模型
原书出处 · 1 处
- · 查看原书依据
three-act narrative archetype, which is common for romantic comedies: “boy meets girl, boy loses girl, boy gets girl back.” For this archetype, they have a corpus of many scripts. In causal terms, the events in act 1 cause the events of act 2, and the events of acts 1 and 2 cause the events of act 3. We can draw the DAG in figure 13.3.
参考来源
- 位置:《Causal AI》第13章《因果与大语言模型》(来源项目:a19c7933-4eda-4176-aebd-30fe7e1ad292;源文件:extracted/Causal AI/_epub-src/OEBPS/Text/chapter-13.xhtml)
- 结论依据:摘录中关于因果脚手架分解文本机制的定义、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:three-act narrative archetype, which is common for romantic comedies: “boy meets girl, boy loses girl, boy gets girl back.” For this archetype, they have a corpus of many scripts. In causal terms, the events in act 1 cause the events of act 2, and the events of acts 1 and 2 cause the events of act 3. We can draw the DAG in figure 13.3.