Skip to content
Code for the paper "Extreme Adaptation for Personalized Neural Machine Translation"
Branch: master
Clone or download
Latest commit 0be061a May 7, 2018
Permalink
Type Name Latest commit message Commit time
Failed to load latest commit information.
config Add files May 4, 2018
scripts Add files May 4, 2018
LICENSE Initial commit May 3, 2018
README.md
attention.py
beam.py
bo_ngrams.py
data.py
decoders.py
encoders.py
evaluation.py
helpers.py
lex.py Add files May 4, 2018
lm.py
make_lex.py
multi_user.py
new_users.py Add files May 4, 2018
options.py Add files May 4, 2018
run.py Add files May 4, 2018
seq2seq.py Add files May 4, 2018
user.py Add files May 4, 2018
utils.py

README.md

Extreme Adaptation for Personalized Neural Machine Translation

This repository contains the code for the paper Extreme Adaptation for Personalized Neural Machine Translation.

Data

The data used in the paper is the SATED dataset, available at this url: http://www.cs.cmu.edu/~pmichel1/sated/.

Additional experiments were performed on the gender annotated europarl corpus from [1], available at this url https://www.kaggle.com/ellarabi/europarl-annotated-for-speaker-gender-and-age.

You can download all the data by running:

# SATED
wget http://www.cs.cmu.edu/~pmichel1/hosting/sated-release-0.9.0.tar.gz
tar xvzf sated-release-0.9.0.tar.gz
# Europarl
wget https://www.kaggle.com/ellarabi/europarl-annotated-for-speaker-gender-and-age/downloads/europarl-annotated-for-speaker-gender-and-age.zip
unzip europarl-annotated-for-speaker-gender-and-age.zip

Requirements

This project was coded in Dynet. It should be working with the 2.0.3 release which you can install by running:

pip install dynet==2.0.3

References

If you use this code or the SATED dataset in you research, consider citing the original paper:

TBD

Other references:

[1]: https://aclanthology.coli.uni-saarland.de/pdf/E/E17/E17-1101.pdf

You can’t perform that action at this time.