ImportError: No module named loader

[AmitMY]

Thanks for this code.

First, the README is missing two dependencies: matplotlib and scipy.

When running the command from the README, I get an error:

Traceback (most recent call last):
File "lattice.py", line 2, in
import msa_class
File "/home/nlp/amit/pseudo-ref/msa_class.py", line 15, in
import loader
ImportError: No module named loader

I am assuming that that is because of

sys.path.append('../pytorch-examples/word_language_model/')

But I haven't found any indication in https://github.com/pytorch/examples/tree/master/word_language_model of a loader.

Can you please add instructions on how to get loader? or maybe add it to this repo as well?

[renj]

Thank you Amit for pointing out this bug. loader actually is in language_model. I've just fixed this.

[AmitMY]

Great, thanks!
One more dependency missing is nltk.

Also, it now fails to import Model:

========= READ DATA COMPLETE ===========
Traceback (most recent call last):
File "lattice.py", line 607, in
lm_model = torch.load(f, map_location=lambda storage, loc: storage)
File "/home/nlp/amit/anaconda2/envs/pseudo-ref/lib/python2.7/site-packages/torch/serialization.py", line 358, in load
return _load(f, map_location, pickle_module)
File "/home/nlp/amit/anaconda2/envs/pseudo-ref/lib/python2.7/site-packages/torch/serialization.py", line 542, in _load
result = unpickler.load()
ImportError: No module named Model

[renj]

It's fixed now. language_model should be added into path.

[AmitMY]

Thanks. This works! I am running the soft code overnight, as it seems to take seconds for some sentences, but many minutes for others, but I ran it on a sample and it worked.

[renj]

Thanks. This works! I am running the soft code overnight, as it seems to take seconds for some sentences, but many minutes for others, but I ran it on a sample and it worked.

Yes. If the lattice is complex, it will take a long time to generate the pseudo-sentences. Actually you can do some optimization on it, e.g. cache the suffixes of each nodes instead of simple DFS.