mosesdecoder/scripts/training/bilingual-lm/README
2014-11-26 10:32:37 +00:00

10 lines
1.1 KiB
Plaintext

Example usage:
#create training and test corpus
/home/abmayne/code/deepathon/nnjm/extract_training.py --working-dir /home/abmayne/experiments/2014-iwslt/nplm/en-cs10k --corpus corpus/europarl.clean.10k --target-language cs --source-language en --align corpus/europarl.clean.10k.align
/home/abmayne/code/deepathon/nnjm/extract_test.py --working-dir /home/abmayne/experiments/2014-iwslt/nplm/en-cs10k --corpus corpus/europarl.test.10k --target-language cs --source-language en --align corpus/europarl.test.10k.align
#Train and test different language models with basic nplm training
/home/abmayne/code/deepathon/nnjm/train_nplm.py --working-dir /home/abmayne/experiments/2014-iwslt/nplm/en-cs10k --corpus europarl.clean.10k --minibatch-size 128 --epochs 40 --output-model europarl.10k.bbn --nplm-home /home/abmayne/code/deepathon/nplm_one_layer --hidden 0 --threads 1 --output-model europarl.10k.1layer
/home/abmayne/code/deepathon/nnjm/test_nplm.py --working-dir /home/abmayne/experiments/2014-iwslt/nplm/en-cs10k --corpus europarl.test.10k --train-corpus europarl.10k.1layer --nplm-home /home/abmayne/code/deepathon/nplm_one_layer --threads 1