JASS

KO | EN

gitlite — search

#javascript #react #ai #llm #android #machine-learning #rust #deep-learning #nodejs #macos #java #cli #pytorch #openai

(★ 16)

JASS: Japanese-specific Sequence to Sequence Pre-training for Neural Machine Translation (LREC2020) & Linguistically Driven Multi-Task Pre-Training for Low-Resource Neural Machine Translation (ACM TALLIP)

JASS 최신버젼 다운로드

최종 버전 다운로드 (.zip)

// repository documentation

Was this content helpful?

★ 0 (0 ratings)

# JASS: Japanese-specific Sequence to Sequence Pre-training for Neural Machine Translation

This the repository for this [paper](https://www.aclweb.org/anthology/2020.lrec-1.454/).

Find extensions of this work and new pre-trained models here: [code](https://github.com/Mao-KU/JASS/tree/master/linguistically-driven-pretraining), [paper](https://arxiv.org/pdf/2201.08070.pdf)

## Requirements
Install OpenNMT-py (1.0) and subword-nmt.
```bash
pip install OpenNMT-py
pip install subword-nmt
```

## Pre-trained JASS models
We release JASS models on 2 language pairs: ja+en, ja+ru. For Japanese seq2seq pretraining, we use our proposed JASS methods while MASS is utilized for English and Russian.

|Model|Vocabulary|BPE codes|
|:---:|:---:|:---:|
|[JASS-jaen](http://lotus.kuee.kyoto-u.ac.jp/~zhuoyuanmao/jaen/model_step_80000.pt)|[ja-en](http://lotus.kuee.kyoto-u.ac.jp/~zhuoyuanmao/jaen/vocab)|[ja-en.bpe.codes](http://lotus.kuee.kyoto-u.ac.jp/~zhuoyuanmao/jaen/joint_bpe.codes)|
|[JASS-jaru](http://lotus.kuee.kyoto-u.ac.jp/~zhuoyuanmao/jaru/model_step_92500.pt)|[ja-ru](http://lotus.kuee.kyoto-u.ac.jp/~zhuoyuanmao/jaru/vocab)|[ja-ru.bpe.codes](http://lotus.kuee.kyoto-u.ac.jp/~zhuoyuanmao/jaru/joint_bpe.codes)|

## Usage
Run the bpe precrocessing for the dataset to be finetuned. After setting up the downloaded vocabulary for src and tgt sentences during the preprocessing phase by ```preprocess.py``` of OpenNMT, use ```train_from``` argument of ```train.py``` in OpenNMT to implement the finetuning for the pretrained model.

## Others
We will update the current Japanese--English pre-trained model and release pretrained models on Japanese--Chinese and Japanese--Korean. We released new models here: [code](https://github.com/Mao-KU/JASS/tree/master/linguistically-driven-pretraining)

## Reference
[1] Zhuoyuan Mao, Fabien Cromieres, Raj Dabre, Haiyue Song, Sadao Kurohashi, [*JASS: Japanese-specific Sequence to Sequence Pre-training for Neural Machine Translation*](https://www.aclweb.org/anthology/2020.lrec-1.454/)

```
@inproceedings{mao-etal-2020-jass,
    title = "{JASS}: {J}apanese-specific Sequence to Sequence Pre-training for Neural Machine Translation",
    author = "Mao, Zhuoyuan  and
      Cromieres, Fabien  and
      Dabre, Raj  and
      Song, Haiyue  and
      Kurohashi, Sadao",
    booktitle = "Proceedings of The 12th Language Resources and Evaluation Conference",
    month = may,
    year = "2020",
    address = "Marseille, France",
    publisher = "European Language Resources Association",
    url = "https://www.aclweb.org/anthology/2020.lrec-1.454",
    pages = "3683--3691",
    language = "English",
    ISBN = "979-10-95546-34-4",
}
```

JASS

JASS 최신버젼 다운로드

오늘의 추천 프로젝트

Was this content helpful?

Recent Feedback

JASS

JASS 최신버젼 다운로드

오늘의 추천 프로젝트

Was this content helpful?

Recent Feedback

Download README