torch.masked_select和torch.masked_scatter

最新推荐文章于 2025-04-04 11:13:11 发布

撒旦即可

最新推荐文章于 2025-04-04 11:13:11 发布

阅读量3.5k

点赞数

分类专栏： Pytorch 文章标签： pytorch 深度学习机器学习

本文链接：https://blog.youkuaiyun.com/qq_39861441/article/details/122694837

版权

Pytorch 专栏收录该内容

36 篇文章

订阅专栏

torch.masked_select

import torch

x = torch.randn(3,4)
print(x)
mask = torch.randn(3,1) >0
print(mask)
s=torch.masked_select(x,mask)
print(s)

输出

tensor([[-2.7601, -0.9957, -0.2239, -1.2674],
[ 1.2190, 1.6782, -0.3635, 0.6803],
[-1.4541, -2.6175, 0.5655, 1.8493]])
tensor([[ True],
[False],
[ True]])
tensor([-2.7601, -0.9957, -0.2239, -1.2674, -1.4541, -2.6175, 0.5655, 1.8493])

import torch

x = torch.randn(3,4)
print(x)
mask = torch.randn(3,1) >0
print(mask)
re=torch.randn(3,4)+100
print(re)
x.masked_scatter_(mask,re)
print(x)

输出

tensor([[-0.2986, -2.1390, -1.3152, 0.1101],
[-0.3291, -0.1943, 0.5282, -0.9538],
[-0.1157, 1.1725, -1.1309, 1.1864]])
tensor([[ True],
[False],
[ True]])
tensor([[100.2276, 99.5916, 100.3636, 98.6109],
[ 98.6181, 98.8684, 100.3451, 100.8166],
[100.1912, 100.7507, 100.6337, 100.2147]])
tensor([[100.2276, 99.5916, 100.3636, 98.6109],
[ -0.3291, -0.1943, 0.5282, -0.9538],
[ 98.6181, 98.8684, 100.3451, 100.8166]])