marcintustin / xnmt

eXtensible Neural Machine Translation

This version

This version is specifically to replicate the work at https://github.com/neulab/word-embeddings-for-nmt

That paper says that the authors used commit 38044b3 but doing so results in this traceback:

   initialized DefaultTranslator({'attender': <xnmt.attender.MlpAttender object at 0x146005208>, 'decoder': <xnmt.decoder.MlpSoftmaxDecoder object at 0x146005048>, 'encoder': <xnmt.lstm.LSTMSeqTransducer object at 0x146005128>, 'src_embedder': <xnmt.embedder.PretrainedSimpleWordEmbedder object at 0x1460050f0>, 'trg_embedder': <xnmt.embedder.PretrainedSimpleWordEmbedder object at 0x146005320>})   
   Traceback (most recent call last):
     File "/usr/local/Cellar/python/3.7.1/Frameworks/Python.framework/Versions/3.7/lib/python3.7/runpy.py", line 193, in _run_module_as_main
          "__main__", mod_spec)   
     File "/usr/local/Cellar/python/3.7.1/Frameworks/Python.framework/Versions/3.7/lib/python3.7/runpy.py", line 85, in _run_code
          exec(code, run_globals)   
     File "/Users/marcin/Downloads/replication/xnmtvenv/lib/python3.7/site-packages/xnmt-0.0.1-py3.7-macosx-10.12-x86_64.egg/xnmt/xnmt_run_experiments.py", line 170, in <module>
          sys.exit(main())   
     File "/Users/marcin/Downloads/replication/xnmtvenv/lib/python3.7/site-packages/xnmt-0.0.1-py3.7-macosx-10.12-x86_64.egg/xnmt/xnmt_run_experiments.py", line 131, in main
          xnmt_trainer.run_epoch()   
     File "/Users/marcin/Downloads/replication/xnmtvenv/lib/python3.7/site-packages/xnmt-0.0.1-py3.7-macosx-10.12-x86_64.egg/xnmt/xnmt_train.py", line 229, in run_epoch
          standard_loss = self.model.calc_loss(src, trg)   
     File "/Users/marcin/Downloads/replication/xnmtvenv/lib/python3.7/site-packages/xnmt-0.0.1-py3.7-macosx-10.12-x86_64.egg/xnmt/translator.py", line 113, in calc_loss
          embeddings = self.src_embedder.embed_sent(src)   
     File "/Users/marcin/Downloads/replication/xnmtvenv/lib/python3.7/site-packages/xnmt-0.0.1-py3.7-macosx-10.12-x86_64.egg/xnmt/embedder.py", line 109, in embed_sent
          embeddings.append(self.embed(batch))   
     File "/Users/marcin/Downloads/replication/xnmtvenv/lib/python3.7/site-packages/xnmt-0.0.1-py3.7-macosx-10.12-x86_64.egg/xnmt/embedder.py", line 87, in embed
          if self.fix_norm != None:   
   AttributeError   :    'PretrainedSimpleWordEmbedder' object has no attribute 'fix_norm'   

The change in this commit fixes that error, and it is presumed that this fix was also applied in the version used in that paper.

eXtensible Neural Machine Translation

This is a repository for the extensible neural machine translation toolkit xnmt. The over-arching goal of xnmt is that it be easy to use for research, and thus it supports a modular design that means that new methods should be easy to implement by adding new modules. It is coded in Python based on DyNet.

More information can be found in the documentation.