知识卡片

普通读书笔记卡 · 6714 · 生成模型:GPT-6

卷积序列长度随配置变化

理解等级:中级

输入长度L、核宽K、步长1且无填充时,输出为L−K+1,而不是固定长度。步长、填充与膨胀还会改变结果;固定宽度文本表示通常还需要全局池化或其他汇聚,不能把卷积本身等同完整编码。

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

      Step 2 is to slide your window across the text and do the dot product of step 1 again. This kernel window sliding, multiplying, and summing is called convolution. Convolutions turn one sequence of numbers into another sequence of numbers that’s about the same size as the original text sequence. Depending on how you do this sliding and multiplying (convolution), you can end up with

    参考来源

    - 位置:《Natural Language Processing in Action, Second Edition》第07章《卷积网络识别文本模式》(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-7.html) - 结论依据:摘录中关于卷积序列长度随配置变化的定义、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:Step 2 is to slide your window across the text and do the dot product of step 1 again. This kernel window sliding, multiplying, and summing is called convolution. Convolutions turn one sequence of numbers into another sequence of numbers that’s about the same size as the original text sequence. Depending on how you do this sliding and multiplying (convolution), you can end up with