Cuda in Line 277 (main.lua)

createmomo commented 5 years ago

Many thanks for the previous quick reply.

There is one more little problem.

Is it possible to adjust the inference step to also support CPU? Because when I used the trained model to tag unlabelled texts, there is an error in the 277 line saying the cuda() is calling.

Many thanks!

ketranm commented 5 years ago

that is actually very simple, you can just comment it out -- input = input:cuda() :)

createmomo commented 5 years ago

Yeah, it works. But it seems there is another problem.

This is my command for tagging. th main.lua -datapath ./data/ -nstates 45 -niters 20 -hidsize 512 -mnbz 256 -nloops 6 -maxlen 81 -nlayers 3 -model ../save/noextra.iter19.t7 -input ./data/train.txt -output -tagged_file.txt

Please find the error here.

vocabulary size: 5382 create networks use Feed-forward Emission Model load model: ../save/noextra.iter19.t7 /home/*/torch/install/bin/luajit: ./Emission.lua:38: bad argument #2 to 'index' (torch.LongTensor expected, got torch.IntTensor) stack traceback: [C]: in function 'index' ./Emission.lua:38: in function 'log_prob' main.lua:282: in function 'infer' main.lua:301: in main chunk [C]: in function 'dofile' ...olin/torch/install/lib/luarocks/rocks/trepl/scm-1/bin/th:150: in main chunk [C]: at 0x00406670

ketranm commented 5 years ago

fixed! Hope it works now.

createmomo commented 5 years ago

Hi ketranm, sorry for my late reply and thank you very much for your help!

ketranm / neuralHMM

Cuda in Line 277 (main.lua) #3

Please find the error here.