pytorch 的numpy.ndarray和tensor和Variable的转换

1. 导入包
 

import torch

import numpy as numpy

from torch.autograd import Variable

2. 定义变量

a = np.array([3, 4, 23])  #numpy类型
type(a)
a_tensor = torch.tensor(a)　　#tensor类型
type(a_tensor)
a_variable = Variable(a_tensor)  # Variable类型
type(a_variable)

3. numpy和tensor之间的转换

# tensor --> numpy
a_t = a_tensor.data # 结果还是tensor
a_t2 = a_tensor.data.numpy() # 结果numpy

# numpy --> tensor
a_t2 = torch.tensor(a_t2)

4. numpy和Variable之间的转换

# Variable  ---->   numpy
a_v = a_variable.data  # 结果为tensor
a_v2 = a_variable.data.numpy()  # 结果为numpy


# numpy --->  Variable 
a_v3 = Variable(a_v2)