PyTorch基本用法(一)——Numpy，Torch对比

文章作者：Tyan
博客：noahsnail.com  |  CSDN  |  简书

本文主要是对比Torch与Numpy的一些操作。

import torchimport numpy as np# numpy的array与torch的tensor的转换np_data = np.arange(6).reshape((2, 3))torch_data = torch.from_numpy(np_data)tensor2array = torch_data.numpy() print 'numpy data: ', np_dataprint 'torch data: ', torch_dataprint 'tensor2array: ', tensor2array

numpy data:  [[0 1 2] [3 4 5]]torch data:   0  1  2 3  4  5[torch.LongTensor of size 2x3]tensor2array:  [[0 1 2] [3 4 5]]

# Tensor的文档：http://pytorch.org/docs/master/tensors.htmldata = [-2, -1, 0, 1, 2]float_data = torch.FloatTensor(data)print float_data

-2-1 0 1 2[torch.FloatTensor of size 5]

# abs操作print np.abs(data)print torch.abs(float_data)

[2 1 0 1 2] 2 1 0 1 2[torch.FloatTensor of size 5]

# sin操作print np.sin(data)print torch.sin(float_data)

[-0.90929743 -0.84147098  0.          0.84147098  0.90929743]-0.9093-0.8415 0.0000 0.8415 0.9093[torch.FloatTensor of size 5]

# mean操作print np.mean(data)print torch.mean(float_data)

0.00.0

# 矩阵相乘data = [[1, 2], [3, 4]]tensor = torch.FloatTensor(data)print np.matmul(data, data)# torch.mm不支持广播形式print torch.mm(tensor, tensor)# torch.matmul支持广播形式print torch.matmul(tensor, tensor)

[[ 7 10] [15 22]]  7  10 15  22[torch.FloatTensor of size 2x2]  7  10 15  22[torch.FloatTensor of size 2x2]