https://github.com/pcyin/pytorch_nmt

A neural machine translation model in PyTorch
https://github.com/pcyin/pytorch_nmt

deep-learning machine-learning neural-machine-translation pytorch

Last synced: about 1 month ago
JSON representation

A neural machine translation model in PyTorch

• Host: GitHub
• URL: https://github.com/pcyin/pytorch_nmt
• Owner: pcyin
• Created: 2017-04-11T18:51:26.000Z (over 7 years ago)
• Default Branch: master
• Last Pushed: 2019-07-03T20:11:21.000Z (over 5 years ago)
• Last Synced: 2024-11-16T04:10:58.377Z (about 2 months ago)
• Topics: deep-learning, machine-learning, neural-machine-translation, pytorch
• Language: Python
• Homepage:
• Size: 16.5 MB
• Stars: 118
• Watchers: 7
• Forks: 25
• Open Issues: 0
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

README

        
**A neural machine translation model written in pytorch.**

**For a up-to-date PyTorch implementation of basic vanilla attentional NMT, please refer to [this repo](https://github.com/pcyin/pytorch_basic_nmt)**

With 256-dimensional LSTM hidden size, it achieves a training speed of 14000 words/sec and 26.9 BLEU score on the IWSLT 2014 Germen-English dataset (Ranzato et al., 2015).

## File Structure

* `nmt.py`: main file

* `vocab.py`: script used to generate `.bin` vocabulary file from parallel corpus

* `util.py`: script containing helper functions

* `run_raml_exp.py|test_raml_models.py`: helper scripts to train|test RAML models with different temperature settings (refer to [Norouzi et al., 2016] for details)

## Usage

* Generate Vocabulary Files

```

python vocab.py

```

* Vanilla Maximum Likelihood Training

```

. scripts/run_mle.sh

```

* Reward Augmented Maximum Likelihood Training (Norouzi et al., 2016)

```

. scripts/run_raml.sh

```

* Reinforcement Learning (Coming soon)

## TODO:

* batched decoding as in openNMT

## License

This work is licensed under a Creative Commons Attribution 4.0 International License.