torch tensor去掉1维_pytorch学习 中 torch.squeeze() 和torch.unsqueeze()的用法

torch.squeeze()

这个函数主要对数据的维度进行压缩，去掉维数为1的的维度，比如是一行或者一列这种，一个一行三列（1,3）的数去掉第一个维数为一的维度之后就变成（3）行。squeeze(a)就是将a中所有为1的维度删掉。不为1的维度没有影响。a.squeeze(N) 就是去掉a中指定的维数为一的维度。还有一种形式就是b=torch.squeeze(a，N) a中去掉指定的定的维数为一的维度。

torch.unsqueeze()

这个函数主要是对数据维度进行扩充。给指定位置加上维数为一的维度，比如原本有个三行的数据（3），在0的位置加了一维就变成一行三列（1,3）。a.squeeze(N) 就是在a中指定位置N加上一个维数为1的维度。还有一种形式就是b=torch.squeeze(a，N) a就是在a中指定位置N加上一个维数为1的维度

折叠 torch.squeeze()

a

tensor([[[-0.1313, -1.0998, -1.9624]]]) 
 torch.Size([1, 1, 3]) 

tensor([-0.1313, -1.0998, -1.9624]) 
 torch.Size([3]) 

tensor([[-0.1313, -1.0998, -1.9624]]) 
 torch.Size([1, 3]) 

tensor([[-0.1313, -1.0998, -1.9624]]) 
 torch.Size([1, 3]) 

tensor([[[-0.1313, -1.0998, -1.9624]]]) 
 torch.Size([1, 1, 3]) 

注意，这里只能去掉维数为1的的维度，如果我们随机生成2x3x4的矩阵，则无效

a=torch.randn(2,3,4)
print(a,"n",a.shape,"n")
b=torch.squeeze(a)
print(b,"n",b.shape,"n")
c=torch.squeeze(a,0)
print(c,"n",c.shape,"n")
d=torch.squeeze(a,1)
print(d,"n",d.shape,"n")
e=torch.squeeze(a,2)#如果去掉第三维，则数不够放了，所以直接保留
print(e,"n",e.shape,"n")

tensor([[[-0.3312, -0.3903,  0.3732, -0.0094],
         [-1.2595,  0.7815, -0.5044,  0.4635],
         [ 0.3063,  0.8799, -1.4904, -1.1514]],

        [[-0.4506, -1.0506, -2.0797, -0.3425],
         [ 1.9772, -0.4648,  0.2649,  0.8535],
         [-0.4897, -0.4739, -0.4632, -0.4432]]]) 
 torch.Size([2, 3, 4]) 

tensor([[[-0.3312, -0.3903,  0.3732, -0.0094],
         [-1.2595,  0.7815, -0.5044,  0.4635],
         [ 0.3063,  0.8799, -1.4904, -1.1514]],

        [[-0.4506, -1.0506, -2.0797, -0.3425],
         [ 1.9772, -0.4648,  0.2649,  0.8535],
         [-0.4897, -0.4739, -0.4632, -0.4432]]]) 
 torch.Size([2, 3, 4]) 

tensor([[[-0.3312, -0.3903,  0.3732, -0.0094],
         [-1.2595,  0.7815, -0.5044,  0.4635],
         [ 0.3063,  0.8799, -1.4904, -1.1514]],

        [[-0.4506, -1.0506, -2.0797, -0.3425],
         [ 1.9772, -0.4648,  0.2649,  0.8535],
         [-0.4897, -0.4739, -0.4632, -0.4432]]]) 
 torch.Size([2, 3, 4]) 

tensor([[[-0.3312, -0.3903,  0.3732, -0.0094],
         [-1.2595,  0.7815, -0.5044,  0.4635],
         [ 0.3063,  0.8799, -1.4904, -1.1514]],

        [[-0.4506, -1.0506, -2.0797, -0.3425],
         [ 1.9772, -0.4648,  0.2649,  0.8535],
         [-0.4897, -0.4739, -0.4632, -0.4432]]]) 
 torch.Size([2, 3, 4]) 

tensor([[[-0.3312, -0.3903,  0.3732, -0.0094],
         [-1.2595,  0.7815, -0.5044,  0.4635],
         [ 0.3063,  0.8799, -1.4904, -1.1514]],

        [[-0.4506, -1.0506, -2.0797, -0.3425],
         [ 1.9772, -0.4648,  0.2649,  0.8535],
         [-0.4897, -0.4739, -0.4632, -0.4432]]]) 
 torch.Size([2, 3, 4]) 

展开 torch.unsqueeze()

a=torch.randn(1,3)
print(a,"n",a.shape,"n")
b=torch.unsqueeze(a,0)
print(b,"n",b.shape,"n")
c=torch.unsqueeze(a,1)
print(c,"n",c.shape,"n")
d=torch.unsqueeze(a,2)
print(d,"n",d.shape,"n")

tensor([[-0.3519, -0.3158,  1.1978]]) 
 torch.Size([1, 3]) 

tensor([[[-0.3519, -0.3158,  1.1978]]]) 
 torch.Size([1, 1, 3]) 

tensor([[[-0.3519, -0.3158,  1.1978]]]) 
 torch.Size([1, 1, 3]) 

tensor([[[-0.3519],
         [-0.3158],
         [ 1.1978]]]) 
 torch.Size([1, 3, 1])