知识卡片
语言生成与知识查询可以分工
查询模块提供受约束的证据或计算结果,语言模块负责按任务组织表达。拆分职责方便检查在哪一步出错,但生成仍可能误述查询结果;需核对最终回答与证据一致,而不能因模块分开就宣布内容必然正确。
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
原书出处 · 1 处
- · 查看原书依据
Grounding can also benefit your NLP pipeline in other ways. Using a knowledge graph for the reasoning part of your algorithm can free up your language model to do what it does best: generate plausible, grammatical text. This allows you to fine-tune your language model to have the tone that you want, without trying to build a chameleon that pretends to understand and reason about
参考来源
- 位置:《Natural Language Processing in Action, Second Edition》第11章《信息抽取与知识图谱》(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-11.html)
- 结论依据:摘录中关于语言生成与知识查询可以分工的定义、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:Grounding can also benefit your NLP pipeline in other ways. Using a knowledge graph for the reasoning part of your algorithm can free up your language model to do what it does best: generate plausible, grammatical text. This allows you to fine-tune your language model to have the tone that you want, without trying to build a chameleon that pretends to understand and reason about