$ word2vec/bin/./word2vec -train ug.mono -output ug.bin -size 100 -sample 1e-5 -negative 5 -threads 8 -binary 1

$ python build-rare.py --data_in train.src --rare_out 1k.train.src.rare --rare_num 1000
$ python build-rare.py --data_in train.trg --rare_out 1k.train.trg.rare --rare_num 1000

$ ./distance-list ug.bin train.src.rare > train.src.rare.sim 2>/dev/null
$ python clean-sim.py --sim_in train.src.rare.sim --sim_out train.src.rare.clean.sim --mono ug.mono

$ python augment.py --data_src train.src --data_trg train.trg \
    --rare_src train.src.rare --rare_trg train.trg.rare \
    --sim_src train.src.rare.clean.sim --sim_trg train.trg.rare.clean.sim \
    --out_src train.src.aug --out_trg train.trg.aug