• Introduction
• Website
• User Manual
• Documentation
• License
• Citation
• Development Team
• Contact

THUMT is a data-driven machine translation system developed by the Natural Language Processing Group at Tsinghua University.

Machine translation is a natural language processing task that aims to translate natural languages using computers automatically. Recent several years have witnessed the rapid development of end-to-end neural machine translation, which has become the new mainstream method in practical MT systems.

On top of Theano, THUMT is an open-source toolkit for neural machine translation with the following features:

• Attention-based translation model. THUMT implements the standard attention-based encoder-decoder framework for NMT.
• Minimum risk training. Besides standard maximum likelihood estimation (MLE), THUMT also supports minimum risk training (MRT) that aims to find a set of model parameters that minimize the expected loss calculated using evaluation metrics such as BLEU on the training data.
• Exploiting monolingual data. THUMT provides semi-supervised training (SST) for NMT that is capable of exploiting abundant monolingual corpora to improve the learning of both source-to-target and target-to-source NMT models.
• Visualization. To better understand the internal workings of NMT, THUMT features a visualization tool to demonstrate the relevance between each intermediate state and its relevant contextual words.

http://thumt.thunlp.org

This user manual describes how to install and use THUMT.

This documentation provides detailed information about the functions in THUMT.

The source code is dual licensed. Open source licensing is under the BSD-3-Clause, which allows free use for research purposes. For commercial licensing, please email [email protected].

Please cite the following paper:

Jiacheng Zhang, Yanzhuo Ding, Shiqi Shen, Yong Cheng, Maosong Sun, Huanbo Luan, Yang Liu. 2017. THUMT: An Open Source Toolkit for Neural Machine Translation. arXiv:1706.06415.

Project leaders: Maosong Sun, Yang Liu, Huanbo Luan

Project members: Jiacheng Zhang, Yanzhuo Ding, Shiqi Shen, Yong Cheng

If you have questions, suggestions and bug reports, please email [email protected].