知识卡片
步长加大会遗漏局部起点
步长2仅访问每隔一个起点的窗口,可能跳过从奇数位置开始的目标组合。是否遗漏依具体模式位置,不可笼统称固定50%成功;增大步长可减少计算,却需核验任务所需的局部覆盖。
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
Later in the chapter, you will use the terms kernel and stride to talk about your stencil and how you slide it across the text. In this case, your stride was 1, and the kernel size was 2. And for the part-of-speech (POS) vectors, your kernel was designed to handle 7D embedding vectors. Had you used the same kernel size of 2 but stepped it
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:摘录中关于步长加大会遗漏局部起点的定义、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:Later in the chapter, you will use the terms kernel and stride to talk about your stencil and how you slide it across the text. In this case, your stride was 1, and the kernel size was 2. And for the part-of-speech (POS) vectors, your kernel was designed to handle 7D embedding vectors. Had you used the same kernel size of 2 but stepped it