pytorch中repeat方法

repeat()沿着特定的维度重复这个张量，按照倍数扩充1、x.repeat(a)列数乘以a倍，对x进行横向赋值import torchx = torch.tensor([1,2,3,4])print(x)print(x.shape)xnew = x.repeat(3)# 注意x的维度没有被改变，repeat后的维度仅仅传入xnew中。print(x)print(xnew)print(xnew.s

啊秋秋秋秋

8278人浏览 · 2021-03-01 10:23:20

啊秋秋秋秋 · 2021-03-01 10:23:20 发布

repeat()沿着特定的维度重复这个张量，按照倍数扩充
1、x.repeat(a)
列数乘以a倍，对x进行横向赋值

import torch
x = torch.tensor([1,2,3,4])
print(x)
print(x.shape)
xnew = x.repeat(3)
# 注意x的维度没有被改变，repeat后的维度仅仅传入xnew中。
print(x)
print(xnew)
print(xnew.shape)

![在这里插入图片描述](https://img-blog.csdnimg.cn/20210301100958501.png

2、x.reshape(a,b)
列数先乘以b倍，再行数乘以a倍。即对x先横向复制b倍，再纵向复制a倍

import torch
x = torch.tensor([1,2,3,4])
print(x)
print(x.shape)
xnew_1 = x.repeat(1,3)
xnew_2 = x.repeat(3,1)
xnew_3 = x.repeat(2,3)
print(xnew_1,xnew_1.shape)
print(xnew_2,xnew_2.shape)
print(xnew_3,xnew_3.shape)

在这里插入图片描述
2、x.reshape(a,b,c)
同理，从c到a变化。

import torch
x = torch.tensor([1,2,3,4])
print(x)
print(x.shape)
xnew_1 = x.repeat(1,1,3)
xnew_2 = x.repeat(1,2,3)
xnew_3 = x.repeat(2,1,1)
xnew_4 = x.repeat(2,1,3)
xnew_5 = x.repeat(2,2,3)
print(xnew_1,xnew_1.shape)
print(xnew_2,xnew_2.shape)
print(xnew_3,xnew_3.shape)
print(xnew_4,xnew_4.shape)
print(xnew_5,xnew_5.shape)