知识卡片

普通读书笔记卡 · 6617.a · 生成模型:GPT-6

隐私规则须覆盖实际格式变化

理解等级:中级

固定长度、有限字符集或少数分隔符的规则可能漏掉其他合法写法,也可能匹配普通数字与文本。应按目标数据构造正反样本并报告范围;书中展示的邮箱和号码正则只是示例,不能直接当完整检测器。

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

      email_pattern = r"[\w\.=-]+ @ [\w\.-]+ \. [\w]{2,3}" Exercise Use the search function in the dataset viewer for the RefinedWeb pre-training dataset to assess presence of PII. For example, search for “gmail.com.” What do you find? Note Removing structured PII data while keeping the number of false positives low is hard enough, but detecting and remediating unstructured data is even harder. Due to the complexity of

    参考来源

    - 位置:《Designing Large Language Model Applications》第02章《预训练数据》(来源项目:a19c7933-4eda-4176-aebd-30fe7e1ad292;源文件:extracted/Designing Large Language Model Applications/_epub-src/OEBPS/ch02.html) - 结论依据:摘录中关于隐私规则须覆盖实际格式变化的定义、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:email_pattern = r"[\w\.=-]+ @ [\w\.-]+ \. [\w]{2,3}" Exercise Use the search function in the dataset viewer for the RefinedWeb pre-training dataset to assess presence of PII. For example, search for “gmail.com.” What do you find? Note Removing structured PII data while keeping the number of false positives low is hard enough, but detecting and remediating unstructured data is even harder. Due to the complexity of