Torch 和 NumPy 对比

NumPy和Torch对比

1. NumPy中ndarray和Torch中tensor的相互转化

   # NumPy中ndarray和Torch中tensor的相互转化
   numpyData = numpy.arange(6).reshape(2, 3)
   # ndarray到tensor
   torchData = torch.from_numpy(numpyData)
   # tensor到ndarray
   tensor2array = torchData.numpy()
   print(
   "\nnumpy array:\n", numpyData,
   "\ntorch tensor:\n", torchData,
   "\ntensor to array\n", tensor2array
   )

运行结果

numpy array:
[[0 1 2]
[3 4 5]] 
torch tensor:
tensor([[0, 1, 2],
[3, 4, 5]]) 
tensor to array
[[0 1 2]
[3 4 5]]

1. 数学运算

• 绝对值

  # 绝对值计算abs
  data = [-1, -2, 1, 2]
  tensor = torch.FloatTensor(data) # 32位
  print(
  '\nabs',
  '\nnumpy:\n', numpy.abs(data),
  '\ntorch:\n', torch.abs(tensor)
  )

运行结果

abs 
numpy:
[1 2 1 2] 
torch:
tensor([1., 2., 1., 2.])

• sin

  # 三角函数sin
  print(
  '\nsin',
  '\nnumpy:\n', numpy.sin(data),
  '\ntorch:\n', torch.sin(tensor)
  )

运行结果

sin 
numpy:
[-0.84147098 -0.90929743  0.84147098  0.90929743] 
torch:
tensor([-0.8415, -0.9093,  0.8415,  0.9093])

• 平均值

  # 平均值
  print(
  '\nmean',
  '\nnumpy:\n', numpy.mean(data),
  '\ntorch:\n', torch.mean(tensor)
  )

运行结果

mean 
numpy:
0.0 
torch:
tensor(0.)

1. 矩阵乘法

   # 矩阵乘法
   data = [[1, 2],
        [3, 4]]
   tensor = torch.FloatTensor(data)
   print(
    '\n叉乘',
    '\nnumpy:\n', numpy.matmul(data, data),
    '\ntorch:\n', torch.mm(tensor, tensor)
   )
   print(
    '\n点乘',
    '\nnumpy:\n', numpy.multiply(data, data),
    '\ntorch:\n', torch.mul(tensor, tensor)
   )

运行结果

叉乘 
numpy:
 [[ 7 10]
 [15 22]] 
torch:
 tensor([[ 7., 10.],
        [15., 22.]])
点乘 
numpy:
 [[ 1  4]
 [ 9 16]] 
torch:
 tensor([[ 1.,  4.],
        [ 9., 16.]])