知识卡片
卷积可处理变长但批处理需约定
卷积本身可接受不同序列长度,结合自适应或全局池化也可输出固定宽度。批内通常需填充、分桶或相应处理,截断则会丢内容;不是所有CNN数学上都只能固定输入长,也不必无条件截掉长文。
这张卡片看不懂?
这张卡片已通过人工审查,不再接受失败反馈。
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
CNN models require consistent-length input text so that all the output values within the encoding are at consistent positions within that vector. This ensures the encoding vector your CNN outputs always has the same number of dimensions no matter how long or short your text is. Your goal is to create vector representations of both a single-character string and a whole page of text. Unfortunately,
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:摘录中关于卷积可处理变长但批处理需约定的定义、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:CNN models require consistent-length input text so that all the output values within the encoding are at consistent positions within that vector. This ensures the encoding vector your CNN outputs always has the same number of dimensions no matter how long or short your text is. Your goal is to create vector representations of both a single-character string and a whole page of text. Unfortunately,