知识卡片

普通读书笔记卡 · 6613.c · 生成模型:GPT-6

启发式清洗是近似筛选

理解等级:中级

句末标点、行长、重复字符与功能词比例可快速过滤部分网页噪声,但合法标题、代码、短句或方言也可能触发。阈值必须结合语料抽样检查,记录误删和漏删,不把启发式命中当成质量真值。

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

      Once text is extracted, the documents are passed through a series of data filtering steps. First, rudimentary filtering steps based on heuristics are applied. While the details differ across datasets, here are some of the steps typically performed:

    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:摘录中关于启发式清洗是近似筛选的定义、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:Once text is extracted, the documents are passed through a series of data filtering steps. First, rudimentary filtering steps based on heuristics are applied. While the details differ across datasets, here are some of the steps typically performed: