This is a Pytorch port of OpenNMT, an open-source (MIT) neural machine translation system.
wget https://s3.amazonaws.com/pytorch/examples/opennmt/data/onmt-data.tar && tar -xf onmt-data.tar
python preprocess.py -train_src data/src-train.txt -train_tgt data/tgt-train.txt -valid_src data/src-val.txt -valid_tgt data/tgt-val.txt -save_data data/demo
python train.py -data data/demo-train.pt -save_model demo_model -gpus 0
python translate.py -gpu 0 -model demo_model_e13_*.pt -src data/src-test.txt -tgt data/tgt-test.txt -replace_unk -verbose -output demo_pred.txt
wget https://raw.githubusercontent.com/moses-smt/mosesdecoder/master/scripts/generic/multi-bleu.perl
perl multi-bleu.perl data/tgt-test.txt < demo_pred.txtData might not come as clean as the demo data. Here is a second example that uses the Moses tokenizer (http://www.statmt.org/moses/) to prepare the Multi30k data from the WMT'16 Multimodal Translation task (http://www.statmt.org/wmt16/multimodal-task.html).
mkdir -p data/multi30k
wget http://www.quest.dcs.shef.ac.uk/wmt16_files_mmt/training.tar.gz && tar -xf training.tar.gz -C data/multi30k && rm training.tar.gz
wget http://www.quest.dcs.shef.ac.uk/wmt16_files_mmt/validation.tar.gz && tar -xf validation.tar.gz -C data/multi30k && rm validation.tar.gz
wget https://staff.fnwi.uva.nl/d.elliott/wmt16/mmt16_task1_test.tgz && tar -xf mmt16_task1_test.tgz -C data/multi30k && rm mmt16_task1_test.tgzwget https://raw.githubusercontent.com/moses-smt/mosesdecoder/master/scripts/tokenizer/tokenizer.perl
wget https://github.com/moses-smt/mosesdecoder/blob/master/scripts/share/nonbreaking_prefixes/nonbreaking_prefix.de
wget https://github.com/moses-smt/mosesdecoder/blob/master/scripts/share/nonbreaking_prefixes/nonbreaking_prefix.en
sed -i "s/$RealBin\/..\/share\/nonbreaking_prefixes//" tokenizer.perl
for l in en de; do for f in data/multi30k/*.$l; do if [[ "$f" != *"test"* ]]; then sed -i "$ d" $f; fi; done; done
for l in en de; do for f in data/multi30k/*.$l; do perl tokenizer.perl -a -no-escape -l $l -q < $f > $f.atok; done; done
python preprocess.py -train_src data/multi30k/train.en.atok -train_tgt data/multi30k/train.de.atok -valid_src data/multi30k/val.en.atok -valid_tgt data/multi30k/val.de.atok -save_data data/multi30k.atok.low -lowerpython train.py -data data/multi30k.atok.low.train.pt -save_model multi30k_model -gpus 0
python translate.py -gpu 0 -model multi30k_model_e13_*.pt -src data/multi30k/test.en.atok -tgt data/multi30k/test.de.atok -replace_unk -verbose -output multi30k.test.pred.atokwget https://raw.githubusercontent.com/moses-smt/mosesdecoder/master/scripts/generic/multi-bleu.perl
perl multi-bleu.perl data/multi30k/test.de.atok < multi30k.test.pred.atokThe following pretrained models can be downloaded and used with translate.py.
- onmt_model_en_de_200k: An English-German translation model based on the 200k sentence dataset at OpenNMT/IntegrationTesting. Perplexity: 21.
- onmt_model_en_fr_b1M: An English-French model trained on benchmark-1M. Perplexity: 4.85.
The following OpenNMT features are implemented:
- multi-layer bidirectional RNNs with attention and dropout
- data preprocessing
- saving and loading from checkpoints
- inference (translation) with batching and beam search
Not yet implemented:
- word features
- multi-GPU
- residual connections
With default parameters on a single Maxwell GPU, this version runs about 70% faster than the Lua torch OpenNMT. The improved performance comes from two main sources:
- CuDNN is used for the encoder (although not for the decoder, since it can't handle attention)
- The decoder softmax layer is batched to efficiently trade off CPU vs. memory efficiency; this can be tuned with the -max_generator_batches parameter.
React
Typescript
Django
Laravel
Facebook
Microsoft
Google
Alibaba
D3
Tencent