In [1]:

    

import torch
from torch.autograd import Variable
import torch.nn as nn
import torch.nn.functional as F


    

In [2]:

    

m = nn.Conv2d(1, 6, 5)


    

In [3]:

    

m


    

    
Out[3]:





Conv2d(1, 6, kernel_size=(5, 5), stride=(1, 1))

In [32]:

    

input = torch.autograd.Variable(torch.randn(1, 1, 32, 32))


    

In [34]:

    

output = m(input)


    

In [35]:

    

output.size()


    

    
Out[35]:





torch.Size([1, 6, 28, 28])

In [36]:

    

m1 = nn.Linear(20,30)


    

In [37]:

    

m1


    

    
Out[37]:





Linear (20 -> 30)

In [38]:

    

input1 = Variable(torch.randn(128,20))


    

In [40]:

    

output1 = m1(input1)


    

In [41]:

    

output1.size()


    

    
Out[41]:





torch.Size([128, 30])

In [49]:

    

x = Variable(torch.randn(2, 2, 2))


    

In [52]:

    

x.view


    

    
Out[52]:





Variable containing:
-0.5078  0.3558  1.3433 -1.9786  0.0182  0.6772 -0.1476 -0.2288
[torch.FloatTensor of size 1x8]

In [50]:

    

x


    

    
Out[50]:





Variable containing:
(0 ,.,.) = 
 -0.5078  0.3558
  1.3433 -1.9786

(1 ,.,.) = 
  0.0182  0.6772
 -0.1476 -0.2288
[torch.FloatTensor of size 2x2x2]

In [55]:

    

torch.arange(1, 11, 2)


    

    
Out[55]:





 1
 3
 5
 7
 9
[torch.FloatTensor of size 5]

In [57]:

    

import torchvision


    

In [ ]: