ParallelCriterion input form?

[jmfu95]

I use : local nll = nn.ClassNLLCriterion() local kl = nn.DistKLDivCriterion() local criterion = nn.ParallelCriterion():add(nll,0.5):add(kl)

and input: {1 : DoubleTensor - size: 50x24 2 : DoubleTensor - size: 50x24}

but i get the following error: torch/install/share/lua/5.1/nn/LookupTable.lua:59: attempt to call method 'isContiguous' (a nil value) stack traceback: ...shedongyu/torch/install/share/lua/5.1/nn/LookupTable.lua:59: in function 'makeInputContiguous' ...shedongyu/torch/install/share/lua/5.1/nn/LookupTable.lua:72: in function 'func' ...hedongyu/torch/install/share/lua/5.1/nngraph/gmodule.lua:345: in function 'neteval' ...hedongyu/torch/install/share/lua/5.1/nngraph/gmodule.lua:380: in function 'forward' ./trainer.lua:76: in function 'opfunc' ...shedongyu/torch/install/share/lua/5.1/optim/adadelta.lua:31: in function 'optim_method' ./trainer.lua:102: in function 'train' main.lua:206: in function 'train_loop' main.lua:393: in function 'main' main.lua:406: in main chunk [C]: in function 'dofile' ...ngyu/torch/install/lib/luarocks/rocks/trepl/scm-1/bin/th:150: in main chunk [C]: at 0x00406670 so i want to know the input form is wrong?

[tastyminerals]

The error means that your input tensor is nil (does not have isContiguous method). Check what you feed into your model.

[jmfu95]

Thanks for replying! It may require input as ParallelTable, so i change the input. But the mistake still exists. So do you know how i should change the input to feed into model?

[tastyminerals]

There is nothing wrong with your parallel container architecture. The error you're are getting means that one of your input DoubleTensors is nil, you are not reading anything from your training data.

[jmfu95]

Thank you! I have fixed this problem.