知识卡片

普通读书笔记卡 · 6676.e · 生成模型:GPT-6

训练高分仍要和独立样本比较

理解等级:中级

在已用于拟合的词向量上得到99.9%准确率,不能证明面对新评论同样有效。独立测试可揭示过拟合;分类之前的分词统计、词表和主题拟合也应纳入训练边界,不能只在最后分类器步骤才切分。

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

      This is a trick question. You have probably already figured out the trap. The reason for this near-perfect 99.9% result is that we haven’t separated out a test set. This A+ score is on “questions” that the classifier has already “seen.” This is like getting an exam in school with the exact same questions you studied the day before. So this model probably wouldn’t do

    参考来源

    - 位置:《Natural Language Processing in Action, Second Edition》第04章《从词数到语义分析》(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-4.html) - 结论依据:摘录中关于训练高分仍要和独立样本比较的定义、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:This is a trick question. You have probably already figured out the trap. The reason for this near-perfect 99.9% result is that we haven’t separated out a test set. This A+ score is on “questions” that the classifier has already “seen.” This is like getting an exam in school with the exact same questions you studied the day before. So this model probably wouldn’t do