← 返回刚才阅读的位置

知识卡片

普通读书笔记卡 · 6792.a · 生成模型:GPT-6

频次优先还要考虑错误影响

理解等级:中级

用户日志可帮助排序常见需求,减少凭内部意见猜测。罕见但影响大的错误仍值得优先处理;高频不等于最重要,新增分支也需验证是否真正改善任务,而不是只降低表面回退率。

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

      Throughout this book, you’ve learned about the importance of human feedback to help train your NLP models to get smarter and smarter over time. You can increase your chatbot’s breadth of knowledge by adding new branches to the dialog tree. And you can increase a chatbot’s ability to understand what your users are saying by finding and labeling utterances your chatbot misunderstood. Figure 12.6 shows

    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:摘录中关于频次优先还要考虑错误影响的定义、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:Throughout this book, you’ve learned about the importance of human feedback to help train your NLP models to get smarter and smarter over time. You can increase your chatbot’s breadth of knowledge by adding new branches to the dialog tree. And you can increase a chatbot’s ability to understand what your users are saying by finding and labeling utterances your chatbot misunderstood. Figure 12.6 shows