← 返回刚才阅读的位置

知识卡片

普通读书笔记卡 · 6752.a · 生成模型:GPT-6

独立测试不应并回训练池随意重用

理解等级:中级

已有测试集承担最终验收时,应保留其角色;若重新划分,应明确原协议已改变,并检查标注可用性与重叠。先随机切出一个集合再反复早停及调参,实质用作验证,不是仍独立的最终测试。

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

      You can download the “Toxic Comment Classification Challenge” dataset (archive.zip) from kaggle.com.35 You can put the data in your $HOME/.nlpia2-data/ directory with all the other large datasets from the book, if you like. When you unzip the archive.zip file, you’ll see it contains the training set (train.csv) and test set (test.csv) as separate CSV files. In the real world, you would probably combine the training

    参考来源

    - 位置:《Natural Language Processing in Action, Second Edition》第09章《可堆叠的Transformer》(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-9.html) - 结论依据:摘录中关于独立测试不应并回训练池随意重用的定义、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:You can download the “Toxic Comment Classification Challenge” dataset (archive.zip) from kaggle.com.35 You can put the data in your $HOME/.nlpia2-data/ directory with all the other large datasets from the book, if you like. When you unzip the archive.zip file, you’ll see it contains the training set (train.csv) and test set (test.csv) as separate CSV files. In the real world, you would probably combine the training