A Neural Machine Translation implementation in PyTorch that also serves as a template for building Seq2Seq models.
./
├── checkpoint
├── chr_en_data
│ ├── dev.chr
│ ├── dev.en
│ ├── test.chr
│ ├── test.en
│ ├── train.chr
│ └── train.en
├── dataset
│ ├── dataset.py
├── decode.py
├── models
│ ├── lstm_seq2seq.py
├── pytest.ini
├── README.md
├── results
├── run.py
├── scripts
│ ├── decode.sh
│ ├── generate_vocab.sh
│ └── run.sh
├── src.model
├── src.vocab
├── tests
│ ├── __init__.py
│ ├── test_bleu.py
│ ├── test_dataset.py
│ ├── test.ipynb
│ ├── test_model.py
│ └── test_vocabEntry.py
├── tgt.model
├── tgt.vocab
├── utils.py
├── vocab.json
└── vocab.py
- generate sentence piece model
./scripts/generate_vocab.sh- train
./scripts/run.sh- decode
./scripts/decode.sh- CS224N: Natural Language Processing with Deep Learning: Stanford course taught by Christopher Manning and Richard Socher. The project structure and organization draw inspiration from the assignments and materials covered in this course.