知识卡片

普通读书笔记卡 · 6717.a · 生成模型:GPT-6

卷积会在输入通道上求和

理解等级:中级

标准groups=1的一个输出通道在所有输入通道和核宽上共同加权求和,再加偏置。四词性通道三位置核有12个权重,输出通道数可独立选择;它不是每个嵌入维度必有一个互不混合的独立输出核。

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

      In the previous example, you had only one dimension to worry about: the adverb tag. Now, you’ll need to work with all four dimensions of these word vectors to get the pattern right. You will also need to coordinate four different “features,” or channels, of data—so for a three-word, four-channel kernel, you will need a 4 × 3 matrix. Each row represents a channel (POS

    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:摘录中关于卷积会在输入通道上求和的定义、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:In the previous example, you had only one dimension to worry about: the adverb tag. Now, you’ll need to work with all four dimensions of these word vectors to get the pattern right. You will also need to coordinate four different “features,” or channels, of data—so for a three-word, four-channel kernel, you will need a 4 × 3 matrix. Each row represents a channel (POS