attempt to call method 'random' (a nil value)

[jKhurana]

Below is my code

require 'dpnn'
 require 'cunn'
 
 local deviceNumber = tonumber(os.getenv("CUDA_CARD_NR"))
 cutorch.setDevice(deviceNumber)
 
 local module = nn.Sequential():cuda()
 module:add(nn.Linear(2,1):cuda())
 module:add(nn.Sigmoid():cuda())
 
 criterion = nn.BCECriterion():cuda() -- Binary Cross Entorpy Criteria
 
 local targets = torch.CudaTensor(10):random(0,1)
 local inputs = torch.CudaTensor(10,2):uniform(-1,1)
 
 function trainEpoch(module,criterion,inputs,targets)
   for i=1,inputs:size(1) do
     local idx = math.random(1,inputs:size(1))
     local input, target = inputs[idx], targets:narrow(1,idx,1)
     -- forward
     local output= module:forward(input)
     local loss= criterion:forward(output,target)
     -- backward
     local gradOutput = criterion:backward(output,target)
     module:zeroGradParameters()
     local gradInput = module:backward(input,gradOutput)
     --update
     module:updateGradParameters(0.9) -- momentum
     module:updateParameters(0.1) -- W = W -0.1*dL/dW
   end
 end
 
 for i=1,100 do
   trainEpoch(module,criterion,inputs,targets)
 end

I am running above using the following command

`CUDA_CARD_NR=1 luajit feedforwad.lua 

`
It gives the following error

    luajit: feedforwad.lua:13: attempt to call method 'random' (a nil value)
    stack traceback:
    	feedforwad.lua:13: in main chunk
    	[C]: at 0x004064f0

I know that there is some error in the line

`local targets = torch.CudaTensor(10):random(0,1)`

But I am not able to figure out. Any help would be appreciated.
Thanks in advance.

[Germanunkol]

Hi,
I've only used torch.Tensor, nt CudaTensor.
I found this issue where they mention that the random API is not fully implemented in CudaTorch.
Can you try to use targets = torch.Tensor( .. ) instead, and later transfer it to the GPU by running targets:cuda()? That's how I've done training on the GPU.
Just as a note: None of this has anything to do with trAInsported code, though. You'll probably be getting more helpful comments on torch forums.

P.S. Training the trains with torch - awesome idea! I look forward to seeing the results! :)