pytorch-创建tensor

Import from numpy

import torch
import numpy as np
a = np.array([2, 3.3])
a

array([2. , 3.3])

a = np.ones([2,3])
torch.from_numpy(a)

tensor([[1., 1., 1.],
        [1., 1., 1.]], dtype=torch.float64)

Import from List

torch.tensor([2., 3.2]) # tensor只接受一个位置参数

tensor([2.0000, 3.2000])

torch.FloatTensor([2., 3.2])

tensor([2.0000, 3.2000])

torch.tensor([[2., 3.2], [1., 22.3300]])

tensor([[ 2.0000,  3.2000],
        [ 1.0000, 22.3300]])

torch.FloatTensor([1, 2])

tensor([1., 2.])

torch.tensor([1, 2])

tensor([1, 2])

uninitialized

a = torch.empty(1)
print(a)
print(a.dim())
print(a.shape[0])
print(a.shape[-1]) # a.size(0)

tensor([0.])
1
1

a=torch.empty(1,2)
print(a)
print(a.dim())
print(a.shape[0])
print(a.shape[1])

tensor([[1.4013e-45, 0.0000e+00]])
2
1
2

a=torch.empty(2,2)
print(a)
print(a.dim())
print(a.shape[0])
print(a.shape[1])

tensor([[-3.7422e+25,  4.5915e-41],
        [ 0.0000e+00,  0.0000e+00]])
2
2
2

torch.Tensor(2, 3)

tensor([[4.2498e+21, 1.3039e-11, 1.4013e-45],
        [0.0000e+00, 1.4013e-45, 0.0000e+00]])

torch.IntTensor(2, 3)

tensor([[1684169014, 1664573795, 1630943078],
        [1664496741, 1647732274, 1681155427]], dtype=torch.int32)

torch.FloatTensor(2,3)

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

set default type

torch.tensor([1.2, 3]).type()

'torch.FloatTensor'

torch.set_default_tensor_type(torch.DoubleTensor) # 设置tensor默认数据类型

torch.tensor([1.2, 3]).type()

'torch.DoubleTensor'

rand/rand_like, randint

torch.rand(3,3)

tensor([[0.3042, 0.4359, 0.3267],
        [0.1789, 0.7789, 0.4212],
        [0.7396, 0.2619, 0.1755]])

a=torch.rand(3,3)
a

tensor([[0.5903, 0.7928, 0.5014],
        [0.6092, 0.0908, 0.6695],
        [0.2991, 0.9321, 0.1164]])

torch.rand_like(a)

tensor([[0.3394, 0.7368, 0.7235],
        [0.5825, 0.9184, 0.0246],
        [0.3461, 0.0210, 0.1008]])

a = torch.randint(1, 10, [2,3])
print(a)
print(a.dim())
print(a.shape[0])
print(a.shape[1])
print("*"*7)
print(a.shape[-1])
print(a.shape[-2])

tensor([[2, 5, 2],
        [6, 8, 9]])
2
2
3
*******
3
2

randn N(0, 1) N(u, std)

torch.randn(3,3)

tensor([[ 1.8307, -0.5371,  1.4034],
        [-1.7542, -1.1900,  0.6335],
        [-0.4314,  0.0911,  0.2340]])

std = torch.arange(1,0,-0.1)
print(std)

tensor([1.0000, 0.9000, 0.8000, 0.7000, 0.6000, 0.5000, 0.4000, 0.3000, 0.2000,
        0.1000])

mean=torch.full([10], 0.0)
print(mean)

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

torch.normal(mean=torch.full([10],0), std=torch.arange(1, 0, -0.1))

---------------------------------------------------------------------------

RuntimeError                              Traceback (most recent call last)

<ipython-input-72-3d01f0f32e3a> in <module>
----> 1 torch.normal(mean=torch.full([10],0), std=torch.arange(1, 0, -0.1))


RuntimeError: Providing a bool or integral fill value without setting the optional `dtype` or `out` arguments is currently unsupported. In PyTorch 1.7, when `dtype` and `out` are not set a bool fill value will return a tensor of torch.bool dtype, and an integral fill value will return a tensor of torch.long dtype.

torch.normal(mean=torch.full([10],0.0), std=torch.arange(1, 0, -0.1))

tensor([-0.1347,  1.0939, -0.3618,  1.3744,  0.8857,  0.4688, -0.5456,  0.0343,
         0.0085, -0.2109])

full

torch.full([2,3], 7.0)  # 填充满

tensor([[7., 7., 7.],
        [7., 7., 7.]])

torch.full([], 7.0) # 设置标量

tensor(7.)

torch.full([1], 7.0)

tensor([7.])

arange/range 推荐使用arange

torch.arange(0,10)

tensor([0, 1, 2, 3, 4, 5, 6, 7, 8, 9])

torch.arange(1,10,2)

tensor([1, 3, 5, 7, 9])

torch.range(0,10)

C:\Anaconda3\envs\pytorchTest\lib\site-packages\ipykernel_launcher.py:1: UserWarning: torch.range is deprecated in favor of torch.arange and will be removed in 0.5. Note that arange generates values in [start; end), not [start; end].
  """Entry point for launching an IPython kernel.





tensor([ 0.,  1.,  2.,  3.,  4.,  5.,  6.,  7.,  8.,  9., 10.])

linspace/logspace

torch.linspace(0, 10, steps = 4) # 等分

tensor([ 0.0000,  3.3333,  6.6667, 10.0000])

torch.linspace(0, 10, steps=10)

tensor([ 0.0000,  1.1111,  2.2222,  3.3333,  4.4444,  5.5556,  6.6667,  7.7778,
         8.8889, 10.0000])

torch.linspace(0, 10, steps=11)

tensor([ 0.,  1.,  2.,  3.,  4.,  5.,  6.,  7.,  8.,  9., 10.])

torch.logspace(0, -1, steps=10) # 10^x

tensor([1.0000, 0.7743, 0.5995, 0.4642, 0.3594, 0.2783, 0.2154, 0.1668, 0.1292,
        0.1000])

torch.logspace(-1, 0, steps=10)

tensor([0.1000, 0.1292, 0.1668, 0.2154, 0.2783, 0.3594, 0.4642, 0.5995, 0.7743,
        1.0000])

10**(-1)

0.1

torch.logspace(0,1,steps=10)

tensor([ 1.0000,  1.2915,  1.6681,  2.1544,  2.7826,  3.5938,  4.6416,  5.9948,
         7.7426, 10.0000])

Ones/zeros/eye

torch.ones(3,3)

tensor([[1., 1., 1.],
        [1., 1., 1.],
        [1., 1., 1.]])

torch.zeros(3,3)

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

a = torch.eye(3,4)
a

tensor([[1., 0., 0., 0.],
        [0., 1., 0., 0.],
        [0., 0., 1., 0.]])

torch.zeros(3)

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

torch.eye(3)

tensor([[1., 0., 0.],
        [0., 1., 0.],
        [0., 0., 1.]])

torch.ones_like(a)

tensor([[1., 1., 1., 1.],
        [1., 1., 1., 1.],
        [1., 1., 1., 1.]])

randperm

num = torch.randperm(10)
num

tensor([4, 1, 5, 7, 3, 9, 8, 6, 0, 2])

a = torch.rand(2,3)
a

tensor([[0.4931, 0.6169, 0.2774],
        [0.8489, 0.4528, 0.6767]])

b = torch.rand(2, 2)
b

tensor([[0.5017, 0.9038],
        [0.5845, 0.4016]])

idx = torch.randperm(2) # 维度随机
print(idx)
a[idx]

tensor([0, 1])





tensor([[0.4931, 0.6169, 0.2774],
        [0.8489, 0.4528, 0.6767]])

idx = torch.randperm(2) # 维度随机
print(idx)
b[idx]

tensor([1, 0])





tensor([[0.5845, 0.4016],
        [0.5017, 0.9038]])

a,b

(tensor([[0.4931, 0.6169, 0.2774],
         [0.8489, 0.4528, 0.6767]]),
 tensor([[0.5017, 0.9038],
         [0.5845, 0.4016]]))