Pytorch创建Tensor常用方法总结

Pytorch创建Tensor常用方法总结

1.从numpy的ndarray或python的list创建
  • torch.from_numpy(ndarray)
  • torch.tensor(list)
  • torch.Tensor(size)
a = np.array([1,2,3])
data = torch.from_numpy(a)
print(data)
"""
输出:
tensor([1, 2, 3], dtype=torch.int32)
"""
b = np.ones([2,3])
data1 = torch.from_numpy(b)
print(data1)
"""
输出:
tensor([[1., 1., 1.],
        [1., 1., 1.]], dtype=torch.float64)
"""
# 参数为 list
print(torch.tensor([2.,1.2]))
"""
输出:
tensor([2.0000, 1.2000])
"""
2.创建未初始化的Tensor
  • torch.empty(size)
  • torch.Tensor数据类型(data)
# 未初始化
data = torch.empty(1)
print(data)
print(torch.Tensor(2,3))
print(torch.IntTensor(3,4))
"""
输出:
tensor([0.])
tensor([[0.0000e+00, 0.0000e+00, 2.8026e-45],
        [0.0000e+00, 1.4013e-45, 0.0000e+00]])
tensor([[1718379891, 1698963500, 1701013878, 1986356256],
        [ 744842089, 1633899296, 1416782188,  543518841],
        [1887007844, 1646275685,  543977327, 1601073006]], dtype=torch.int32)
"""
 
print(torch.tensor([1.,2]).type())
torch.set_default_tensor_type(torch.DoubleTensor)
print(torch.tensor([1.,2]).type())
"""
输出:
torch.FloatTensor
torch.DoubleTensor
"""
3.随机生成
  • torch.rand(size) 产生[0,1] 均匀分布的数据
  • torch.rand_like(input, dtype) 接收tensor读取shape再用rand生成
  • torch.randint(low = 0, high, size) 随机生成整数值tensor,范围 [min,max) 左闭右开
  • torch.randn(size) N(0,1)均值为0,方差为1的正态分布(N(u,std))
  • torch.full(size, fill_value) 全部赋予相同的值
  • torch.normal(means,std,out = None) 返回一个张量,包含从给定参数means,std的离散正态分布中抽取随机数。 均值means是一个张量,包含每个输出元素相关的正态分布的均值。 std是一个张量,包含每个输出元素相关的正态分布的标准差。 均值和标准差的形状不须匹配,但每个张量的元素个数须相同 参数:means (Tensor) – 均值 std (Tensor) – 标准差 out (Tensor) – 可选的输出张量
data = torch.rand(3,3)
print(data)
"""
输出:
tensor([[0.0775, 0.2610, 0.0833],
        [0.7911, 0.6999, 0.6589],
        [0.4790, 0.6801, 0.6582]])
"""
data_like = torch.randn_like(data)
print(data_like)
"""
输出:
tensor([[ 0.6866,  2.5939, -0.2480],
        [-0.9259, -0.3617,  0.5759],
        [-1.0179, -1.0938,  0.6426]])
"""
print(torch.randint(1,10,[3,3]))
"""
输出:
tensor([[7, 3, 2],
        [8, 6, 7],
        [7, 7, 7]])
"""
 
data = torch.randn(3,3)
print(data)
"""
输出:
tensor([[-0.6225, -0.1253, -0.1083],
        [-0.3199, -0.5670,  0.2898],
        [-0.6500,  0.9275,  1.0377]])
"""
data = torch.normal(mean=torch.full([10],0),std=torch.arange(1,0,-0.1))
print(data)
"""
输出:
tensor([-0.6509, -1.4877,  0.4740,  1.1891,  0.1009, -0.4449, -0.3422,  0.1519,
        -0.2735,  0.1140])
"""
print(torch.full([2,4],7))
print(torch.full([],7)) # 标量
"""
输出:
tensor([[7., 7., 7., 7.],
        [7., 7., 7., 7.]])
tensor(7.)
"""
4.序列生成
  • torch.arange(start, end, step) [start,end) 左闭右开,默认步长为1
  • torch.range(start, end, step) (已被arange替代) 包括end,step是两个点间距
  • torch.linspace(start, end, steps) 等差数列 包括end, steps 是点的个数,包括端点, (等距离)
  • torch.logspace(start, end, steps)
print(torch.arange(0,10))
print(torch.arange(0,10,2))
print(torch.linspace(0,10,steps=3))
print(torch.linspace(0,10,steps=11))
print(torch.logspace(0,-1,steps=10))
print(torch.logspace(0,1,steps=10))
"""
输出:
tensor([0, 1, 2, 3, 4, 5, 6, 7, 8, 9])
tensor([0, 2, 4, 6, 8])
tensor([ 0.,  5., 10.])
tensor([ 0.,  1.,  2.,  3.,  4.,  5.,  6.,  7.,  8.,  9., 10.])
tensor([1.0000, 0.7743, 0.5995, 0.4642, 0.3594, 0.2783, 0.2154, 0.1668, 0.1292,
        0.1000])
tensor([ 1.0000,  1.2915,  1.6681,  2.1544,  2.7826,  3.5938,  4.6416,  5.9948,
         7.7426, 10.0000])
"""
5.全0,全1,单位矩阵
  • torch.zeros(size)
  • torch.zeros_like(input, dtype)
  • torch.ones(size)
  • torch.ones_like(input, dtype)
  • torch.eye(size)
print(torch.ones(3,3))
print(torch.zeros(2,3))
print(torch.eye(4,4))
print(torch.eye(2))
"""
输出:
tensor([[1., 1., 1.],
        [1., 1., 1.],
        [1., 1., 1.]])
tensor([[0., 0., 0.],
        [0., 0., 0.]])
tensor([[1., 0., 0., 0.],
        [0., 1., 0., 0.],
        [0., 0., 1., 0.],
        [0., 0., 0., 1.]])
tensor([[1., 0.],
        [0., 1.]])
"""
6.torch.randperm(n)
  • torch.randperm(n) 生成一个0到n-1的n个整数的随机排列
print(torch.randperm(10))
"""
输出:
tensor([0, 1, 4, 7, 9, 8, 6, 3, 2, 5])
"""

你可能感兴趣的:(Pytorch学习,pytorch,python,深度学习)