知识卡片

普通读书笔记卡 · 6717 · 生成模型:GPT-6

卷积核参数轴不是样本轴

理解等级:中级

Conv1d权重形状为输出通道×输入通道每组数×核宽,新增最前一维表示一个输出滤波器,而非一个样本。输入则通常为批次×输入通道×序列长度;即使两者都三维,各轴职责也不同。

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 PyTorch or any other deep learning framework designed to process multiple samples in parallel, you have to unsqueeze the kernel to add a dimension to hold additional samples. Your unsqueezed kernel (weight matrix) needs to be the same shape as your batch of input data. The first dimension is for the samples from your training or test datasets that are being input to the

    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:摘录中关于卷积核参数轴不是样本轴的定义、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 PyTorch or any other deep learning framework designed to process multiple samples in parallel, you have to unsqueeze the kernel to add a dimension to hold additional samples. Your unsqueezed kernel (weight matrix) needs to be the same shape as your batch of input data. The first dimension is for the samples from your training or test datasets that are being input to the