WHAT IS PYTORCH

pytorch是：
1）可以使用GPU的Numpy
2）深度学习的框架

1.Tensors
Tensors类似于Numpy的ndarrays，区别在于Tensor可以使用GPU。

• 创建5*3的矩阵，未初始化

from __future__ import print_function
import torch
x = torch.empty(5, 3)
print(x)
----------------------------------------------
tensor([[ 5.0375e+28,  4.5625e-41,  5.0195e+28],
        [ 4.5625e-41, -1.8338e+30,  3.0840e-41],
        [-9.9826e+08,  4.5625e-41, -1.5343e+08],
        [ 4.5625e-41, -1.0043e+09,  4.5625e-41],
        [-2.0086e+08,  4.5625e-41, -9.5843e+08]])

• 创建随机初始化的矩阵

x = torch.rand(5, 3)
print(x)
-------------------------------------------------
tensor([[0.5814, 0.0997, 0.1744],
        [0.2834, 0.9581, 0.9954],
        [0.9372, 0.4401, 0.1696],
        [0.1424, 0.5370, 0.9970],
        [0.6686, 0.5558, 0.5354]])

-创建0初始化矩阵，并设定类型为long

x = torch.zeros(5, 3, dtype=torch.long)
print(x)
---------------------------------------------------
tensor([[0, 0, 0],
        [0, 0, 0],
        [0, 0, 0],
        [0, 0, 0],
        [0, 0, 0]])

• 从列表中创建

x = torch.tensor([5.5, 3])
print(x)
---------------------------------------------------
tensor([5.5000, 3.0000])

• 从已存在的tensor中创建,新的tensor将使用旧tensor的属性，除非新创建的指定了新的属性

x = x.new_ones(5, 3, dtype=torch.double)      # new_* methods take in sizes
print(x)

x = torch.randn_like(x, dtype=torch.float)    # override dtype!
print(x)                                      # result has the same size
------------------------------------------------------
tensor([[1., 1., 1.],
        [1., 1., 1.],
        [1., 1., 1.],
        [1., 1., 1.],
        [1., 1., 1.]], dtype=torch.float64)
tensor([[-0.0970,  1.1034, -1.6941],
        [-1.7651, -0.5884, -1.1931],
        [ 1.0376, -0.8236,  0.8907],
        [ 0.4683, -0.1217,  1.2467],
        [ 0.4624,  0.4772, -1.0577]])

2.操作

• 加法

y = torch.rand(5, 3)
print(x + y)
-------------------------------
tensor([[ 0.4675,  1.7053, -1.2332],
        [-1.6031,  0.1028, -0.6090],
        [ 1.8545, -0.6408,  1.0778],
        [ 0.9508,  0.7635,  2.1345],
        [ 0.5693,  1.0205, -0.5476]])
--------------------------------
print(torch.add(x, y))
----------------------------------
tensor([[ 0.4675,  1.7053, -1.2332],
        [-1.6031,  0.1028, -0.6090],
        [ 1.8545, -0.6408,  1.0778],
        [ 0.9508,  0.7635,  2.1345],
        [ 0.5693,  1.0205, -0.5476]])
----------------------------------
# adds x to y
y.add_(x)
print(y)
----------------------------------
tensor([[ 0.4675,  1.7053, -1.2332],
        [-1.6031,  0.1028, -0.6090],
        [ 1.8545, -0.6408,  1.0778],
        [ 0.9508,  0.7635,  2.1345],
        [ 0.5693,  1.0205, -0.5476]])

• indexing

print(x[:, 1])
---------------------------------------------
tensor([ 1.1034, -0.5884, -0.8236, -0.1217,  0.4772])

• resizing

x = torch.randn(4, 4)
y = x.view(16)
z = x.view(-1, 8)  # the size -1 is inferred from other dimensions
print(x.size(), y.size(), z.size())
--------------------------------------------------
torch.Size([4, 4]) torch.Size([16]) torch.Size([2, 8])

• item 获得结果

x = torch.randn(1)
print(x)
print(x.item())
------------------------------------------------------
tensor([0.9551])
0.9551321864128113

3.Numpy桥梁
Torch Tensor和Numpy array会共享内存空间(如果Torch Tensor在CPU)，改变其中一个另一个也会随着改变。

• 将tensor转换为ndarry

a = torch.ones(5)
print(a)
---------------------
tensor([1., 1., 1., 1., 1.])
---------------------
b = a.numpy()
print(b)
---------------------
[1. 1. 1. 1. 1.]
---------------------
a.add_(1)
print(a)
print(b)
---------------------
tensor([2., 2., 2., 2., 2.])
[2. 2. 2. 2. 2.]

• 将ndarry转换为numpy

import numpy as np
a = np.ones(5)
b = torch.from_numpy(a)
np.add(a, 1, out=a)
print(a)
print(b)
---------------------
[2. 2. 2. 2. 2.]
tensor([2., 2., 2., 2., 2.], dtype=torch.float64)

4.CUDA Tensor

# let us run this cell only if CUDA is available
# We will use ``torch.device`` objects to move tensors in and out of GPU
if torch.cuda.is_available():
    device = torch.device("cuda")          # a CUDA device object
    y = torch.ones_like(x, device=device)  # directly create a tensor on GPU
    x = x.to(device)                       # or just use strings ``.to("cuda")``
    z = x + y
    print(z)
    print(z.to("cpu", torch.double))       # ``.to`` can also change dtype together!
---------------------
tensor([1.9551], device='cuda:0')
tensor([1.9551], dtype=torch.float64)

参考：
https://pytorch.org/tutorials/beginner/blitz/tensor_tutorial.html#sphx-glr-beginner-blitz-tensor-tutorial-py