Hello Hakimeh,
the branch https://github.com/moses-smt/mosesdecoder/tree/nmt-hybrid
supports NMT (which is basically an RNN conditioned on the source text)
as a feature function in Moses. It is described in this paper:
@inproceedings{junczysdowmunt-dwojak-sennrich:2016:WMT,
address = "Berlin, Germany",
author = "Junczys-Dowmunt, Marcin and Dwojak, Tomasz and Sennrich,
Rico",
booktitle = "{Proceedings of the First Conference on Machine
Translation}",
month = "August",
pages = "319--325",
publisher = "Association for Computational Linguistics",
title = "{The AMU-UEDIN Submission to the WMT16 News Translation
Task: Attention-based NMT Models as Feature Functions in Phrase-based SMT}",
url = "http://www.aclweb.org/anthology/W/W16/W16-2316",
year = "2016"
}
I'm afraid that documentation is currently lacking though.
best wishes,
Rico
On 08/12/16 08:34, Shafagh Fadaee wrote:
Dear Moses Community,
I'm searching for some RNN-based features added to Moses decoder or used in
reranking Moses' n-best list.
It seems Moses have some (LM) features based on Feedforward neural networks but
I haven't find any implementation for recurrent neural networks in Moses.
Apparently there are many efforts in this field but I have trouble finding
released codes to use as guideline in my work.
I would be happy to hear your suggestions.
Best regards,
Hakimeh Fadaei
_______________________________________________
Moses-support mailing list
Moses-support@mit.edu
http://mailman.mit.edu/mailman/listinfo/moses-support
_______________________________________________
Moses-support mailing list
Moses-support@mit.edu
http://mailman.mit.edu/mailman/listinfo/moses-support