NICT_LOGO.JPG KYOTO-U_LOGO.JPG

WAT

The Workshop on Asian Translation
Evaluation Results

[EVALUATION RESULTS TOP] | [BLEU] | [RIBES] | [AMFM] | [HUMAN (WAT2022)] | [HUMAN (WAT2021)] | [HUMAN (WAT2020)] | [HUMAN (WAT2019)] | [HUMAN (WAT2018)] | [HUMAN (WAT2017)] | [HUMAN (WAT2016)] | [HUMAN (WAT2015)] | [HUMAN (WAT2014)] | [EVALUATION RESULTS USAGE POLICY]

BLEU


# Team Task Date/Time DataID BLEU
Method
Other
Resources
System
Description
juman kytea mecab moses-
tokenizer
stanford-
segmenter-
ctb
stanford-
segmenter-
pku
indic-
tokenizer
unuse myseg kmseg
1Bering LabJPCNko-ja2021/04/25 02:58:43551275.6876.3575.82-------NMTYesTransformer Ensemble with additional crawled parallel corpus
2TMUJPCNko-ja2021/04/14 19:40:06515173.6774.5373.92-------NMTNoJapanese BART, ensemble of 3 models
3TMUJPCNko-ja2020/09/17 22:35:27382973.1374.0373.45-------NMTNoTransformer, domain adaptation (BERT Japanese), shared EMB, shared BPE, ensemble of 4 models
4TMUJPCNko-ja2020/09/17 22:38:16383073.1073.9973.41-------NMTNoTransformer, domain adaptation (BERT Japanese), hanja loss, shared EMB, shared BPE, ensemble of 4 models
5TMUJPCNko-ja2020/10/13 09:38:06413773.0773.9573.38-------NMTNoTransformer, domain adaptation (BERT Korean), hanja loss, shared EMB, shared BPE, ensemble of 4 models
6TMUJPCNko-ja2020/10/13 09:51:57414173.0673.9773.39-------NMTNoTransformer, domain adaptation (BERT Korean), shared EMB, shared BPE, ensemble of 4 models
7KNU_HyundaiJPCNko-ja2019/07/26 13:26:52299873.0473.9573.32-------NMTNoTransformer + Relative Position + ensemble of 11 models
8sarahJPCNko-ja2019/07/25 12:58:28292472.5573.4472.83-------NMTNoTransformer, ensemble of 4 models
9goku20JPCNko-ja2020/09/18 17:17:24391872.2073.0772.48-------NMTNoTransformer, ensemble of 3 models
10ryanJPCNko-ja2019/07/24 15:48:35289071.7872.6772.08-------NMTNoBase Transformer
11EHRJPCNko-ja2018/08/31 18:58:50221471.7072.6872.01----- 0.00 0.00NMTNoSMT reranked NMT
12sarahJPCNko-ja2019/07/22 12:44:02279471.4172.3371.71-------NMTNoTransformer, single model
13ORGANIZERJPCNko-ja2018/08/16 16:21:17199770.1370.9770.39----- 0.00 0.00NMTNoNMT with Attention
14goku20JPCNko-ja2020/09/18 17:28:33392769.3770.4169.68-------NMTNomBART pre-training, ensemble of 3 models
15tpt_watJPCNko-ja2021/04/27 01:54:36569863.4964.4563.88-------NMTNoBase Transformer model with joint vocab, size 8k

Notice:

Back to top

RIBES


# Team Task Date/Time DataID RIBES
Method
Other
Resources
System
Description
juman kytea mecab moses-
tokenizer
stanford-
segmenter-
ctb
stanford-
segmenter-
pku
indic-
tokenizer
unuse myseg kmseg
1Bering LabJPCNko-ja2021/04/25 02:58:4355120.9520250.9517290.951775-------NMTYesTransformer Ensemble with additional crawled parallel corpus
2TMUJPCNko-ja2021/04/14 19:40:0651510.9518220.9512400.951721-------NMTNoJapanese BART, ensemble of 3 models
3TMUJPCNko-ja2020/09/17 22:38:1638300.9507910.9502470.950528-------NMTNoTransformer, domain adaptation (BERT Japanese), hanja loss, shared EMB, shared BPE, ensemble of 4 models
4TMUJPCNko-ja2020/10/13 09:38:0641370.9507210.9501790.950498-------NMTNoTransformer, domain adaptation (BERT Korean), hanja loss, shared EMB, shared BPE, ensemble of 4 models
5TMUJPCNko-ja2020/09/17 22:35:2738290.9505060.9498160.950269-------NMTNoTransformer, domain adaptation (BERT Japanese), shared EMB, shared BPE, ensemble of 4 models
6TMUJPCNko-ja2020/10/13 09:51:5741410.9504010.9497280.950233-------NMTNoTransformer, domain adaptation (BERT Korean), shared EMB, shared BPE, ensemble of 4 models
7sarahJPCNko-ja2019/07/25 12:58:2829240.9498970.9493990.949706-------NMTNoTransformer, ensemble of 4 models
8goku20JPCNko-ja2020/09/18 17:17:2439180.9498380.9491770.949627-------NMTNoTransformer, ensemble of 3 models
9KNU_HyundaiJPCNko-ja2019/07/26 13:26:5229980.9493040.9487090.949060-------NMTNoTransformer + Relative Position + ensemble of 11 models
10EHRJPCNko-ja2018/08/31 18:58:5022140.9485290.9474680.948416-----0.0000000.000000NMTNoSMT reranked NMT
11ryanJPCNko-ja2019/07/24 15:48:3528900.9479860.9473900.947783-------NMTNoBase Transformer
12goku20JPCNko-ja2020/09/18 17:28:3339270.9473970.9468780.947266-------NMTNomBART pre-training, ensemble of 3 models
13sarahJPCNko-ja2019/07/22 12:44:0227940.9472540.9469950.947183-------NMTNoTransformer, single model
14ORGANIZERJPCNko-ja2018/08/16 16:21:1719970.9419500.9413360.941905-----0.0000000.000000NMTNoNMT with Attention
15tpt_watJPCNko-ja2021/04/27 01:54:3656980.9328050.9312720.931897-------NMTNoBase Transformer model with joint vocab, size 8k

Notice:

Back to top

AMFM


# Team Task Date/Time DataID AMFM
Method
Other
Resources
System
Description
unuse unuse unuse unuse unuse unuse unuse unuse unuse unuse
1TMUJPCNko-ja2021/04/14 19:40:0651510.9359410.9359410.935941-------NMTNoJapanese BART, ensemble of 3 models
2Bering LabJPCNko-ja2021/04/25 02:58:4355120.9303190.9303190.930319-------NMTYesTransformer Ensemble with additional crawled parallel corpus
3tpt_watJPCNko-ja2021/04/27 01:54:3656980.9027310.9027310.902731-------NMTNoBase Transformer model with joint vocab, size 8k
4ORGANIZERJPCNko-ja2018/08/16 16:21:1719970.0000000.0000000.000000-----0.0000000.000000NMTNoNMT with Attention
5EHRJPCNko-ja2018/08/31 18:58:5022140.0000000.0000000.000000-----0.0000000.000000NMTNoSMT reranked NMT
6sarahJPCNko-ja2019/07/22 12:44:0227940.0000000.0000000.000000-------NMTNoTransformer, single model
7ryanJPCNko-ja2019/07/24 15:48:3528900.0000000.0000000.000000-------NMTNoBase Transformer
8sarahJPCNko-ja2019/07/25 12:58:2829240.0000000.0000000.000000-------NMTNoTransformer, ensemble of 4 models
9KNU_HyundaiJPCNko-ja2019/07/26 13:26:5229980.0000000.0000000.000000-------NMTNoTransformer + Relative Position + ensemble of 11 models
10TMUJPCNko-ja2020/09/17 22:35:2738290.0000000.0000000.000000-------NMTNoTransformer, domain adaptation (BERT Japanese), shared EMB, shared BPE, ensemble of 4 models
11TMUJPCNko-ja2020/09/17 22:38:1638300.0000000.0000000.000000-------NMTNoTransformer, domain adaptation (BERT Japanese), hanja loss, shared EMB, shared BPE, ensemble of 4 models
12goku20JPCNko-ja2020/09/18 17:17:2439180.0000000.0000000.000000-------NMTNoTransformer, ensemble of 3 models
13goku20JPCNko-ja2020/09/18 17:28:3339270.0000000.0000000.000000-------NMTNomBART pre-training, ensemble of 3 models
14TMUJPCNko-ja2020/10/13 09:38:0641370.0000000.0000000.000000-------NMTNoTransformer, domain adaptation (BERT Korean), hanja loss, shared EMB, shared BPE, ensemble of 4 models
15TMUJPCNko-ja2020/10/13 09:51:5741410.0000000.0000000.000000-------NMTNoTransformer, domain adaptation (BERT Korean), shared EMB, shared BPE, ensemble of 4 models

Notice:

Back to top

HUMAN (WAT2022)


# Team Task Date/Time DataID HUMAN
Method
Other
Resources
System
Description

Notice:
Back to top

HUMAN (WAT2021)


# Team Task Date/Time DataID HUMAN
Method
Other
Resources
System
Description

Notice:
Back to top

HUMAN (WAT2020)


# Team Task Date/Time DataID HUMAN
Method
Other
Resources
System
Description
1TMUJPCNko-ja2020/09/17 22:35:2738294.710NMTNoTransformer, domain adaptation (BERT Japanese), shared EMB, shared BPE, ensemble of 4 models
2TMUJPCNko-ja2020/09/17 22:38:1638304.700NMTNoTransformer, domain adaptation (BERT Japanese), hanja loss, shared EMB, shared BPE, ensemble of 4 models
3goku20JPCNko-ja2020/09/18 17:28:3339274.640NMTNomBART pre-training, ensemble of 3 models

Notice:
Back to top

HUMAN (WAT2019)


# Team Task Date/Time DataID HUMAN
Method
Other
Resources
System
Description
1sarahJPCNko-ja2019/07/25 12:58:282924UnderwayNMTNoTransformer, ensemble of 4 models
2KNU_HyundaiJPCNko-ja2019/07/26 13:26:522998UnderwayNMTNoTransformer + Relative Position + ensemble of 11 models

Notice:
Back to top

HUMAN (WAT2018)


# Team Task Date/Time DataID HUMAN
Method
Other
Resources
System
Description

Notice:
Back to top

HUMAN (WAT2017)


# Team Task Date/Time DataID HUMAN
Method
Other
Resources
System
Description

Notice:
Back to top

HUMAN (WAT2016)


# Team Task Date/Time DataID HUMAN
Method
Other
Resources
System
Description

Notice:
Back to top

HUMAN (WAT2015)


# Team Task Date/Time DataID HUMAN
Method
Other
Resources
System
Description

Notice:
Back to top

HUMAN (WAT2014)


# Team Task Date/Time DataID HUMAN
Method
Other
Resources
System
Description

Notice:
Back to top

EVALUATION RESULTS USAGE POLICY

When you use the WAT evaluation results for any purpose such as:
- writing technical papers,
- making presentations about your system,
- advertising your MT system to the customers,
you can use the information about translation directions, scores (including both automatic and human evaluations) and ranks of your system among others. You can also use the scores of the other systems, but you MUST anonymize the other system's names. In addition, you can show the links (URLs) to the WAT evaluation result pages.

NICT (National Institute of Information and Communications Technology)
Kyoto University
Last Modified: 2018-08-02