Pytorch中的torch.nn.functional.conv2d与深度可分离卷积和标准卷积 | 小王同学

https://a171232886.github.io/2021/11/24/Pytorch%E4%B8%ADtorch.nn.functional.conv2d%E4%B8%8E%E6%B7%B1%E5%BA%A6%E5%8F%AF%E5%88%86%E7%A6%BB%E5%8D%B7%E7%A7%AF%E5%92%8C%E6%A0%87%E5%87%86%E5%8D%B7%E7%A7%AF/

前言
是在研究训练过程中遇到的
F.conv2d与nn.Conv2d
也就是torch.nn.functional.conv2d实际上是torch.nn.Conv2d的另一个用法，可以直接指定卷积核和偏置的值。 这在通常情况下是用不到的，因为卷积核的值都是训练得到的。但在一些相关运算（correlation）时，需要指定卷积核的值。首先回顾一下nn.Conv2d
# Conv2d的用法 i

The issue links to a blog post about PyTorch's conv2d functions. Review the post to understand the explanation of torch.nn.functional.conv2d vs nn.Conv2d and depthwise separable convolution. The task likely involves updating or clarifying documentation on the site. Check the repository's structure for where such content lives, possibly in _posts or a similar directory. Ensure the changes align with the site's formatting and style.

Vom Indexierungsmodell aus dem Issue-Text verfasst.