使用PyTorch进行数学运算

import torch
x = torch.zeros((3, 2))
y = torch.ones((3, 2))
z = torch.rand((3, 2))
print(x)
print(y)
print(z)

tensor([[0., 0.],
        [0., 0.],
        [0., 0.]])
tensor([[1., 1.],
        [1., 1.],
        [1., 1.]])
tensor([[0.9255, 0.9734],
        [0.3468, 0.9825],
        [0.3859, 0.7481]])

import torch

a = torch.rand((5, 3))

print(a)  # 全部
print(a[0])  # 第1行
print(a[0][2])  # 第1行第3列
print(a[1, 2])  # 第2行第3列
print(a[:2])  # 前3行
print(a[:, 1])  # 第2列

tensor([[0.9835, 0.6304, 0.3341], 
        [0.1006, 0.8246, 0.2030], 
        [0.7149, 0.7477, 0.7167], 
        [0.7997, 0.5218, 0.9917], 
        [0.9539, 0.2919, 0.0582]])
tensor([0.9835, 0.6304, 0.3341])
tensor(0.3341)
tensor(0.2030)
tensor([[0.9835, 0.6304, 0.3341],
        [0.1006, 0.8246, 0.2030]])
tensor([0.6304, 0.8246, 0.7477, 0.5218, 0.2919])

import torch
x = torch.ones((2, 2)) * 2
y = torch.ones((2, 2))

a = x.add(y)  # x + y
b = x.sub(y)  # x - y
c = x.mul(y)  # x * y，按元素乘法
d = x.div(y)  # x / y
e = x.mm(y)  # 矩阵乘法
f = x.t()  # 矩阵的逆

print(a)
print(b)
print(c)
print(d)
print(e)
print(f)

tensor([[3., 3.],
        [3., 3.]])
tensor([[1., 1.],
        [1., 1.]])
tensor([[2., 2.],
        [2., 2.]])
tensor([[2., 2.],
        [2., 2.]])
tensor([[4., 4.],
        [4., 4.]])
tensor([[2., 2.],
        [2., 2.]])

import torch
import numpy as np

x_tensor = torch.randn(2, 3)
y_numpy = np.random.randn(2, 3)

x_numpy = x_tensor.numpy()
y_tensor = torch.from_numpy(y_numpy)

z_tensor = torch.FloatTensor(y_numpy)

print(x_numpy)
print(y_tensor)
print(z_tensor)

[[-1.2588831   0.57915723 -1.0662982 ]
 [-0.0852809   0.13274558  0.19468519]]
tensor([[ 1.4440,  2.3114,  0.6984],
        [ 0.3031, -0.2362,  0.9197]], dtype=torch.float64)
tensor([[ 1.4440,  2.3114,  0.6984],
        [ 0.3031, -0.2362,  0.9197]])

import torch

x = torch.rand(2, 3)
y = torch.randn(2, 3)

cuda = torch.cuda.is_available()
print("cuda available: %s" % cuda)

a = x.cuda()
b = y.cuda()
c = a + b
print(c)

a = a.cpu()
b = b.cpu()
c = a + b
print(c)

cuda available: True
tensor([[ 0.5795,  0.7328,  1.2017],
        [ 1.3393,  0.9051, -0.1900]], device='cuda:0')
tensor([[ 0.5795,  0.7328,  1.2017],
        [ 1.3393,  0.9051, -0.1900]])

import torch

x = torch.ones(2, 2, requires_grad=True)
print(x.data)

y = x + 2
print(y.data)
print(y.grad_fn)

z = y * y
print(z.data)
print(z.grad_fn)

t = torch.mean(z)  # 每个元素求和 / 元素个数
print(t.data)

t.backward()
print(z.grad)
print(y.grad)
print(x.grad)

tensor([[1., 1.],
        [1., 1.]])
tensor([[3., 3.],
        [3., 3.]])
<AddBackward0 object at 0x0000027485E81DC0>
tensor([[9., 9.],
        [9., 9.]])
<MulBackward0 object at 0x0000027485E81DC0>
tensor(9.)
d:\Project\LearnPyTorch\sample01\02.py:18: UserWarning: The .grad attribute of a Tensor that is not a leaf Tensor is being accessed. Its .grad attribute won't be populated during autograd.backward(). If you indeed want the .grad field to be populated for a non-leaf Tensor, use .retain_grad() on the non-leaf Tensor. If you access the non-leaf Tensor by mistake, make sure you access the leaf Tensor instead. See github.com/pytorch/pytorch/pull/30531 for more informations. (Triggered internally at C:\actions-runner\_work\pytorch\pytorch\builder\windows\pytorch\build\aten\src\ATen/core/TensorBody.h:485.)
  print(z.grad)
None
d:\Project\LearnPyTorch\sample01\02.py:19: UserWarning: The .grad attribute of a Tensor that is not a leaf Tensor is being accessed. Its .grad attribute won't be populated during autograd.backward(). If you indeed want the .grad field to be populated for a non-leaf Tensor, use .retain_grad() on the non-leaf Tensor. If you access the non-leaf Tensor by mistake, make sure you access the leaf Tensor instead. See github.com/pytorch/pytorch/pull/30531 for more informations. (Triggered internally at C:\actions-runner\_work\pytorch\pytorch\builder\windows\pytorch\build\aten\src\ATen/core/TensorBody.h:485.)
  print(y.grad)
None
tensor([[1.5000, 1.5000],
        [1.5000, 1.5000]])