notebook.community

Edit and run



In [2]:

    
require 'nngraph';



In [8]:

    
a = torch.Tensor{1,2,3}
print(a)









    Out[8]:





 1
 2
 3
[torch.DoubleTensor of size 3]

For whatever reason, the identity module is the first step of every nngraph



In [6]:

    
module1 = nn.Identity()



In [9]:

    
module1:forward(a)









    Out[9]:





 1
 2
 3
[torch.DoubleTensor of size 3]

Notice the extra parentheses. "The extra () contain properties of this module when embedded into a graph"



In [10]:

    
x1 = nn.Identity()()
m = nn.gModule({x1},{x1})



In [11]:

    
m:forward(a)









    Out[11]:





 1
 2
 3
[torch.DoubleTensor of size 3]

Digression: the Oxford tutorial:



In [14]:

    
add = nn.CAddTable()
t1 = torch.Tensor{3,4,10}
x=add:forward({a,t1})



In [16]:

    
x









    Out[16]:





  4
  6
 13
[torch.DoubleTensor of size 3]

Creating z = x1 + x1 * linear(x3)



In [10]:

    
-- Declare some tensors
t1 = torch.Tensor{1,2,3}
t2 = torch.Tensor{3,4,5}
x1 = nn.Identity()()
x2 = nn.Identity()()
a = nn.CAddTable()({x1,x2})
m = nn.gModule({x1,x2},{a})
print(m:forward({t1,t2}))









    Out[10]:





 4
 6
 8
[torch.DoubleTensor of size 3]

Back to Lab



In [9]:

    
x1 = nn.Identity()()
x2 = nn.Identity()()
add = nn.CAddTable()({x1,x2})
mul = nn.CMulTable()({add,x1})
m = nn.gModule({x1,x2},{mul})
print(m:forward({t1,t2}))









    



...rlesguthrie/torch/install/share/lua/5.1/nngraph/init.lua:48: inputs[1] should be an nngraph.Node but is of type torch.DoubleTensor
stack traceback:
	[C]: in function 'error'
	...rlesguthrie/torch/install/share/lua/5.1/nngraph/init.lua:48: in function <...rlesguthrie/torch/install/share/lua/5.1/nngraph/init.lua:25>
	[C]: at 0x028340c0
	[string "x1 = nn.Identity()()..."]:3: in main chunk
	[C]: in function 'xpcall'
	...arlesguthrie/torch/install/share/lua/5.1/itorch/main.lua:179: in function <...arlesguthrie/torch/install/share/lua/5.1/itorch/main.lua:143>
	...arlesguthrie/torch/install/share/lua/5.1/lzmq/poller.lua:75: in function 'poll'
	...esguthrie/torch/install/share/lua/5.1/lzmq/impl/loop.lua:307: in function 'poll'
	...esguthrie/torch/install/share/lua/5.1/lzmq/impl/loop.lua:325: in function 'sleep_ex'
	...esguthrie/torch/install/share/lua/5.1/lzmq/impl/loop.lua:370: in function 'start'
	...arlesguthrie/torch/install/share/lua/5.1/itorch/main.lua:350: in main chunk
	[C]: in function 'require'
	(command line):1: in main chunk
	[C]: at 0x010234bbb0



In [36]:

    
a = torch.Tensor{1,2,3,1}
b = torch.Tensor{3,4,5,0}
print(m:forward({a,b}))









    Out[36]:





  4
 12
 24
  1
[torch.DoubleTensor of size 4]

A4

nngraph



In [4]:

    
x = torch.ones(4)
y = torch.ones(5)
z = torch.ones(2)



In [7]:

    
--inputs
ix = nn.Identity()()
iy = nn.Identity()()
iz = nn.Identity()()
--Wx + b
h1 = nn.Linear(4,2)({ix})
h2 = nn.Linear(5,2)({iy})
-- tanh, sigmoid
tanh = nn.Tanh()({h1})
sigmoid = nn.Sigmoid()({h2})
-- square
tsq = nn.Square()({tanh})
ssq = nn.Square()({sigmoid})
-- cmul
cmul = nn.CMulTable()({tsq,ssq})
a = nn.CAddTable()({cmul,iz})
-- final graph
output = nn.gModule({ix,iy,iz},{a})



In [8]:

    
h1.data.module.weight = torch.ones(2,4)
h1.data.module.bias = torch.ones(2)
h2.data.module.weight = torch.ones(2,5)
h2.data.module.bias = torch.ones(2)



In [12]:

    
print(output:forward({x,y,z}))









    Out[12]:





 1.9949
 1.9949
[torch.DoubleTensor of size 2]



In [ ]:

    
graph.dot(output.fg, 'output','outputBaseName')

nngraph b



In [111]:

    
gradOutput = torch.ones(2)



In [112]:

    
foo = {[0]=1,[1]=5}



In [113]:

    
foo









    Out[113]:





{







    Out[113]:





  0 : 1
  1 : 5
}



In [122]:

    
foo = torch.ones(4,4)



In [123]:

    
foo









    Out[123]:





 1  1  1  1
 1  1  1  1
 1  1  1  1
 1  1  1  1
[torch.DoubleTensor of size 4x4]



In [125]:

    
print(nn.Reshape(2,2):forward(foo))









    Out[125]:





(1,.,.) = 
  1  1
  1  1

(2,.,.) = 
  1  1
  1  1

(3,.,.) = 
  1  1
  1  1

(4,.,.) = 
  1  1
  1  1
[torch.DoubleTensor of size 4x2x2]



In [ ]:

    
vm = torch.load('vocab_map.tab')



In [ ]: