Pytorch summary learning series - data manipulation

In deep learning , We usually do data entry frequently ⾏ That's ok 行 operation . Act as ⼿ Hand learning is the foundation of deep learning , This section describes how to do this internally
The data in the memory goes into ⾏ That's ok 行 operation .

stay PyTorch in , torch.Tensor Is the main way to store and transform data ⼯ Tools . If you had ⽤ Used to NumPy, You'll find that
Tensor and NumPy Multidimensional array of ⾮ Very similar . however ⽽, Tensor Provide GPU The calculation and ⾃ Automatic gradient Wait a minute 更 multi-function , this
Some emissaries Tensor more 更 It is suitable for deep learning .

tensor It means tensor , A multidimensional tensor is a multidimensional array , Scalar is 0 D tensor , A vector is a 1 D tensor , A matrix is a two-dimensional tensor .

establish TENSOR

First, import. PyTorch：

import torch 

And then we establish One 5x3 Of Uninitialized Tensor ：

x = torch.empty(5, 3)
print(x)

Output ：

tensor([[ 0.0000e+00, 1.5846e+29, 0.0000e+00],
[ 1.5846e+29, 5.6052e-45, 0.0000e+00],
[ 0.0000e+00, 0.0000e+00, 0.0000e+00],
[ 0.0000e+00, 0.0000e+00, 0.0000e+00],
[ 0.0000e+00, 1.5846e+29, -2.4336e+02]])

establish ⼀ One 5x3 Of Randomly initialized Tensor :

Output ：

tensor([[0.4963, 0.7682, 0.0885],
[0.1320, 0.3074, 0.6341],
[0.4901, 0.8964, 0.4556],
[0.6323, 0.3489, 0.4017],
[0.0223, 0.1689, 0.2939]])

establish ⼀ One 5x3 Of long Type all 0 Of Tensor :

x = torch.zeros(5, 3, dtype=torch.long)
print(x)

Output ：

tensor([[0, 0, 0],
[0, 0, 0],
[0, 0, 0],
[0, 0, 0],
[0, 0, 0]])

Create directly from the data :

x = torch.tensor([5.5, 3])
print(x)

Output ：

tensor([5.5000, 3.0000]) 

just so so Through the existing Tensor To create , this ⽅ Method defaults to redo ⽤ Use transport ⼊ Enter into Tensor Of ⼀ Some properties , example 例 Such as data type , except ⾮ Not
⾃ Custom data types .

x = x.new_ones(5, 3, dtype=torch.float64) # Back to tensor The default has the same
torch.dtype and torch.device
print(x)
x = torch.randn_like(x, dtype=torch.float) # Specify a new data type
print(x)

Output ：

tensor([[1., 1., 1.],
[1., 1., 1.],
[1., 1., 1.],
[1., 1., 1.],
[1., 1., 1.]], dtype=torch.float64)
tensor([[ 0.6035, 0.8110, -0.0451],
[ 0.8797, 1.0482, -0.0445],
[-0.7229, 2.8663, -0.5655],
[ 0.1604, -0.0254, 1.0739],
[ 2.2628, -0.9175, -0.2251]])

adopt shape perhaps size() To get Tensor The shape of the :

print(x.size())
print(x.shape)

Output ：

torch.Size([5, 3])
torch.Size([5, 3])

Be careful ： Back to torch.Size In fact, that is ⼀ One tuple, ⽀ Support all tuple The operation of .

Many functions can be created Tensor , Go to the official ⽅https://pytorch.org/docs/1.2.0/torch.htmlAPI You know the 了, The following table gives 了⼀ Some often ⽤ For reference .

  These create ⽅ Methods can specify the data type when creating dtype And storage device(cpu/gpu).