CS-Net

CS-Net copied to clipboard

Why is it nan when I use CS2Net to calculate it?

chuanqiWen

fix a runtime error in later pytorch versions

1

RuntimeError: one of the variables needed for gradient computation has been modified by an inplace operation

amedyukhina

Line 128 in train.py. Whether the return should cancel indentation?

ShiSan0013

https://insight-journal.org/midas/community/view/21

caijinyue

ChannelAttentionBlock函数部分的代码不能理解，还请作者答疑解惑。 困惑的代码为 affinity = torch.matmul(proj_query, proj_key) affinity_new = torch.max(affinity, -1, keepdim=True)[0].expand_as(affinity) - affinity affinity_new = self.softmax(affinity_new) 为什么要减去affinity得到affinity_new？这样做不就使得affinity低的channel获得更高的权重吗？这不就与原本的目的背道而驰了吗？ 您的文章里也并没有详细说明这一个运算的意义，所以本人无法理解该部分代码。 还请作者指点迷津。

Liusj-gh

您好，请问CS2-NET中提到的OCTA数据集现在公开了吗

1

ExiaSit