In [1]:

    

import torch
from torch.autograd import Variable


    

In [2]:

    

x = Variable(torch.ones(2,2), requires_grad=True)
print(x)


    

    




Variable containing:
 1  1
 1  1
[torch.FloatTensor of size 2x2]

In [4]:

    

y = x + 2
print(y)


    

    




Variable containing:
 3  3
 3  3
[torch.FloatTensor of size 2x2]

In [5]:

    

print(y.grad_fn)


    

    




<torch.autograd.function.AddConstantBackward object at 0x00000253ABF8FBE8>

In [7]:

    

z = y*y*3
out = z.mean()
print(z,out)


    

    




Variable containing:
 27  27
 27  27
[torch.FloatTensor of size 2x2]
 Variable containing:
 27
[torch.FloatTensor of size 1]

In [8]:

    

out.backward()


    

In [9]:

    

print(x.grad)


    

    




Variable containing:
 4.5000  4.5000
 4.5000  4.5000
[torch.FloatTensor of size 2x2]

In [11]:

    

x = torch.randn(3)
x = Variable(x, requires_grad=True)

y = x * 2
while y.data.norm() < 1000:
    y = y * 2
print(y)


    

    




Variable containing:
 -419.2958
  136.8486
-1727.5154
[torch.FloatTensor of size 3]

In [35]:

    

gradients = torch.FloatTensor([0.1, 1.0, 0.0001])
y.backward(gradients)
print(x.grad)


    

    




Variable containing:
  409.6000
 4096.0000
    0.4096
[torch.FloatTensor of size 3]

In [17]:

    

torch.norm  # 范数


    

    
Out[17]:





<function torch._C.norm>

In [16]:

    

y.data.norm(p=10)


    

    
Out[16]:





1727.5155034377074

In [23]:

    

a = torch.randn(4, 2)
print(a)


    

    




-0.2103 -1.2730
-0.4334 -0.3583
-0.5065 -2.1537
 0.2715  1.7556
[torch.FloatTensor of size 4x2]

In [25]:

    

torch.norm(a, 0, 1)


    

    
Out[25]:





 2
 2
 2
 2
[torch.FloatTensor of size 4x1]

In [32]:

    

torch.norm(a, 1, -1)


    

    
Out[32]:





 1.4833
 0.7917
 2.6602
 2.0272
[torch.FloatTensor of size 4x1]

In [51]:

    

x1 = torch.ones(3)
x1 = Variable(x1, requires_grad=True)

y1 = x1 * 2
while y1.data.norm() < 1000:
    y1 = y1 * 2

z = y1 * 10


    

In [48]:

    

gradients = torch.FloatTensor([0.1, 1.0, 0.0001])
z.backward(gradients)

print(x1.grad)


    

    




Variable containing:
  2048.0000
 20480.0000
     2.0480
[torch.FloatTensor of size 3]

In [ ]: