permute()函数

最新推荐文章于 2024-07-12 01:14:57 发布

xulei_zhai@163.com

最新推荐文章于 2024-07-12 01:14:57 发布

阅读量505

点赞数

分类专栏：目标检测

本文链接：https://blog.csdn.net/weixin_45135154/article/details/118445989

版权

python

目标检测专栏收录该内容

12 篇文章 1 订阅

订阅专栏

import torch
a = torch.randn(3,2,4)
b = a.permute(1,2,0)
print(a)
print("a.shape", a.shape)
print("(3,2,4)-->(2,4,3)")
print(b)
print("b.shape", b.shape)

tensor([[[-0.3183,  1.2260,  0.4502,  1.1955],
         [ 0.0197, -0.7851,  0.0231, -1.3826]],

        [[ 0.4614,  0.3729, -0.5784, -0.0914],
         [-0.4267, -1.2463,  0.2844,  0.6883]],

        [[-1.7502, -0.6467, -1.0851,  1.9374],
         [ 0.2634,  0.5507,  0.8899,  1.4475]]])
a.shape--> torch.Size([3, 2, 4])
(3,2,4)-->(2,4,3)
tensor([[[-0.3183,  0.4614, -1.7502],
         [ 1.2260,  0.3729, -0.6467],
         [ 0.4502, -0.5784, -1.0851],
         [ 1.1955, -0.0914,  1.9374]],

        [[ 0.0197, -0.4267,  0.2634],
         [-0.7851, -1.2463,  0.5507],
         [ 0.0231,  0.2844,  0.8899],
         [-1.3826,  0.6883,  1.4475]]])
b.shape--> torch.Size([2, 4, 3])