ketranm / neuralhmm Goto Github PK

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!

Hi,

This is a very interesting project. How is it licensed? Could one of the maintainers add a license to it to make explicit any legal encumbrances for the project?

Thanks!

Hi,

I'm really excited about this project. The corresponding paper outlines exactly some of the goals I have for a specific model. I was glad to find the paper and this repository.

I've been reviewing the code in preparation to write a pytorch implementation. But since I don't want to redo work that someone's already done, I'd like to ask if you're aware of an existing pytorch implementation before I get too deep into writing code.

And if you're not aware of a pytroch implementation of a neuralHMM, would you be interested in collaborating to write one?

Thanks,

Hi,

I am running the model on my own laptop using CPU. However, I encountered the error. Probably it is the compatibility between using CPU and GPU?

(I did not install cudnn.torch, because for now, I don't have GPUs to train)

/home/*/torch/install/bin/luajit: ./BaumWelch.lua:146: invalid arguments: DoubleTensor ByteTensor
expected arguments: DoubleTensor [DoubleTensor] DoubleTensor
stack traceback:
[C]: in function 'cmul'
./BaumWelch.lua:146: in function 'run'
main.lua:160: in function 'train'
main.lua:300: 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

Please find more details below:

Command:
th main.lua -datapath ../wsj -nstates 45 -niters 20 -hidsize 512 -mnbz 256 -nloops 6 -maxlen 81 -nlayers 3 -modelpath ../save -model noextra

Output:
{
hidsize : 512
output : "../data/pred.txt"
cuda : false
max_word_len : 15
model : "noextra"
max_norm : 5
niters : 20
nstates : 45
debug : false
maxlen : 81
kernels :
{
1 : 1
2 : 2
3 : 3
4 : 4
5 : 5
6 : 6
7 : 7
}
nlayers : 3
nloops : 6
feature_maps :
{
1 : 50
2 : 100
3 : 128
4 : 128
5 : 128
6 : 128
7 : 128
}
charsize : 15
report_every : 10
maxchars : 15
modelpath : "../save"
datapath : "../wsj"
input : ""
lstm : false
dropout : 0.5
mnbz : 256
vocabsize : -1
conv : false
}
run preprocessing
use all words!
vocabulary size: 19160
create networks
use Feed-forward Emission Model
/home/*/torch/install/bin/luajit: ./BaumWelch.lua:146: invalid arguments: DoubleTensor ByteTensor
expected arguments: DoubleTensor [DoubleTensor] DoubleTensor
stack traceback:
[C]: in function 'cmul'
./BaumWelch.lua:146: in function 'run'
main.lua:160: in function 'train'
main.lua:300: 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

Hi,

Could you please upload an example of the input file for training the model?

Thank you very much!