WAT

The Workshop on Asian Translation

Evaluation Results

BLEU

#	Team	Task	Date/Time	DataID	BLEU										Method	Other Resources	System Description
#	Team	Task	Date/Time	DataID	juman	kytea	mecab	moses- tokenizer	stanford- segmenter- ctb	stanford- segmenter- pku	indic- tokenizer	unuse	myseg	kmseg	Method	Other Resources	System Description
1	ORGANIZER	JPCN3en-ja	2018/08/15 12:49:22	1937	45.75	47.55	45.87	-	-	-	-	-	0.00	0.00	NMT	No	NMT with Attention
2	EHR	JPCN3en-ja	2018/09/08 20:37:52	2250	50.75	52.43	50.86	-	-	-	-	-	0.00	0.00	NMT	No	SMT reranked NMT
3	EHR	JPCN3en-ja	2018/09/13 12:53:28	2286	50.04	51.81	50.33	-	-	-	-	-	0.00	0.00	NMT	Yes	SMT reranked NMT (4M traning data from WAT and NTCIR, Epoch 13)
4	EHR	JPCN3en-ja	2018/09/15 15:55:42	2397	51.17	52.79	51.48	-	-	-	-	-	0.00	0.00	NMT	Yes	SMT reranked NMT (4M traning data from WAT and NTCIR, Epoch 18)
5	EHR	JPCN3en-ja	2018/09/16 15:12:04	2478	51.42	53.06	51.74	-	-	-	-	-	0.00	0.00	NMT	Yes	SMT reranked NMT (4M traning data from WAT and NTCIR, Epoch 20)
6	sarah	JPCN3en-ja	2019/07/26 11:26:55	2975	49.57	51.12	49.85	-	-	-	-	-	-	-	NMT	No	Transformer, ensemble of 4 models
7	KNU_Hyundai	JPCN3en-ja	2019/07/27 12:09:01	3187	53.28	54.52	53.56	-	-	-	-	-	-	-	NMT	No	Transformer Base, relative position, BT, r2l reranking, checkpoint ensemble
8	KNU_Hyundai	JPCN3en-ja	2019/07/27 12:53:49	3195	54.84	56.06	55.06	-	-	-	-	-	-	-	NMT	Yes	Transformer Base (+ASPEC data), relative position, BT, r2l reranking, checkpoint ensemble
9	goku20	JPCN3en-ja	2020/09/21 12:28:46	4103	49.80	51.34	49.72	-	-	-	-	-	-	-	NMT	No	mBART pre-training transformer, single model
10	goku20	JPCN3en-ja	2020/09/22 00:19:09	4116	50.42	51.99	50.43	-	-	-	-	-	-	-	NMT	No	mBART pre-training transformer, ensemble of 3 models
11	tpt_wat	JPCN3en-ja	2021/04/27 02:29:11	5708	51.22	52.93	51.28	-	-	-	-	-	-	-	NMT	No	Base Transformer model with joint vocab, size 8k
12	Bering Lab	JPCN3en-ja	2021/05/04 20:03:17	6375	53.02	54.26	52.74	-	-	-	-	-	-	-	NMT	Yes	Transformer Ensemble with additional crawled parallel corpus
13	sakura	JPCN3en-ja	2024/08/08 19:53:13	7278	53.50	54.90	53.40	-	-	-	-	-	-	-	NMT	No	LLM: Rakuten/RakutenAI-7B-chat Fine-Tuned with JPC Corpus in six direction (En-Ja, Ja-En, Ko-Ja, Ja-Ko, Zh-Ja, Ja-Zh) - Best
14	sakura	JPCN3en-ja	2024/08/09 00:52:14	7308	52.90	54.30	52.80	-	-	-	-	-	-	-	NMT	No	LLM: Rakuten/RakutenAI-7B-chat Fine-Tuned with JPC Corpus in six direction (En-Ja, Ja-En, Ko-Ja, Ja-Ko, Zh-Ja, Ja-Zh)

Notice:

This table is sorted by the leftmost segmenters. You can change the segmenter used to sort by clicking each segmenter link.

RIBES

#	Team	Task	Date/Time	DataID	RIBES										Method	Other Resources	System Description
#	Team	Task	Date/Time	DataID	juman	kytea	mecab	moses- tokenizer	stanford- segmenter- ctb	stanford- segmenter- pku	indic- tokenizer	unuse	myseg	kmseg	Method	Other Resources	System Description
1	ORGANIZER	JPCN3en-ja	2018/08/15 12:49:22	1937	0.843837	0.843885	0.843641	-	-	-	-	-	0.000000	0.000000	NMT	No	NMT with Attention
2	EHR	JPCN3en-ja	2018/09/08 20:37:52	2250	0.869863	0.868636	0.869047	-	-	-	-	-	0.000000	0.000000	NMT	No	SMT reranked NMT
3	EHR	JPCN3en-ja	2018/09/13 12:53:28	2286	0.864863	0.862649	0.864034	-	-	-	-	-	0.000000	0.000000	NMT	Yes	SMT reranked NMT (4M traning data from WAT and NTCIR, Epoch 13)
4	EHR	JPCN3en-ja	2018/09/15 15:55:42	2397	0.864236	0.863230	0.864157	-	-	-	-	-	0.000000	0.000000	NMT	Yes	SMT reranked NMT (4M traning data from WAT and NTCIR, Epoch 18)
5	EHR	JPCN3en-ja	2018/09/16 15:12:04	2478	0.864493	0.863806	0.864485	-	-	-	-	-	0.000000	0.000000	NMT	Yes	SMT reranked NMT (4M traning data from WAT and NTCIR, Epoch 20)
6	sarah	JPCN3en-ja	2019/07/26 11:26:55	2975	0.860350	0.859735	0.859644	-	-	-	-	-	-	-	NMT	No	Transformer, ensemble of 4 models
7	KNU_Hyundai	JPCN3en-ja	2019/07/27 12:09:01	3187	0.878702	0.877109	0.878338	-	-	-	-	-	-	-	NMT	No	Transformer Base, relative position, BT, r2l reranking, checkpoint ensemble
8	KNU_Hyundai	JPCN3en-ja	2019/07/27 12:53:49	3195	0.883523	0.881559	0.883160	-	-	-	-	-	-	-	NMT	Yes	Transformer Base (+ASPEC data), relative position, BT, r2l reranking, checkpoint ensemble
9	goku20	JPCN3en-ja	2020/09/21 12:28:46	4103	0.861492	0.858869	0.860337	-	-	-	-	-	-	-	NMT	No	mBART pre-training transformer, single model
10	goku20	JPCN3en-ja	2020/09/22 00:19:09	4116	0.867687	0.865149	0.866649	-	-	-	-	-	-	-	NMT	No	mBART pre-training transformer, ensemble of 3 models
11	tpt_wat	JPCN3en-ja	2021/04/27 02:29:11	5708	0.871699	0.870395	0.870765	-	-	-	-	-	-	-	NMT	No	Base Transformer model with joint vocab, size 8k
12	Bering Lab	JPCN3en-ja	2021/05/04 20:03:17	6375	0.876937	0.874509	0.874385	-	-	-	-	-	-	-	NMT	Yes	Transformer Ensemble with additional crawled parallel corpus
13	sakura	JPCN3en-ja	2024/08/08 19:53:13	7278	0.881645	0.879535	0.879699	-	-	-	-	-	-	-	NMT	No	LLM: Rakuten/RakutenAI-7B-chat Fine-Tuned with JPC Corpus in six direction (En-Ja, Ja-En, Ko-Ja, Ja-Ko, Zh-Ja, Ja-Zh) - Best
14	sakura	JPCN3en-ja	2024/08/09 00:52:14	7308	0.880617	0.878024	0.878626	-	-	-	-	-	-	-	NMT	No	LLM: Rakuten/RakutenAI-7B-chat Fine-Tuned with JPC Corpus in six direction (En-Ja, Ja-En, Ko-Ja, Ja-Ko, Zh-Ja, Ja-Zh)

Notice:

This table is sorted by the leftmost segmenters. You can change the segmenter used to sort by clicking each segmenter link.

AMFM

#	Team	Task	Date/Time	DataID	AMFM										Method	Other Resources	System Description
#	Team	Task	Date/Time	DataID	unuse	unuse	unuse	unuse	unuse	unuse	unuse	unuse	unuse	unuse	Method	Other Resources	System Description
1	ORGANIZER	JPCN3en-ja	2018/08/15 12:49:22	1937	0.000000	0.000000	0.000000	-	-	-	-	-	0.000000	0.000000	NMT	No	NMT with Attention
2	EHR	JPCN3en-ja	2018/09/08 20:37:52	2250	0.000000	0.000000	0.000000	-	-	-	-	-	0.000000	0.000000	NMT	No	SMT reranked NMT
3	EHR	JPCN3en-ja	2018/09/13 12:53:28	2286	0.000000	0.000000	0.000000	-	-	-	-	-	0.000000	0.000000	NMT	Yes	SMT reranked NMT (4M traning data from WAT and NTCIR, Epoch 13)
4	EHR	JPCN3en-ja	2018/09/15 15:55:42	2397	0.000000	0.000000	0.000000	-	-	-	-	-	0.000000	0.000000	NMT	Yes	SMT reranked NMT (4M traning data from WAT and NTCIR, Epoch 18)
5	EHR	JPCN3en-ja	2018/09/16 15:12:04	2478	0.000000	0.000000	0.000000	-	-	-	-	-	0.000000	0.000000	NMT	Yes	SMT reranked NMT (4M traning data from WAT and NTCIR, Epoch 20)
6	sarah	JPCN3en-ja	2019/07/26 11:26:55	2975	0.000000	0.000000	0.000000	-	-	-	-	-	-	-	NMT	No	Transformer, ensemble of 4 models
7	KNU_Hyundai	JPCN3en-ja	2019/07/27 12:09:01	3187	0.000000	0.000000	0.000000	-	-	-	-	-	-	-	NMT	No	Transformer Base, relative position, BT, r2l reranking, checkpoint ensemble
8	KNU_Hyundai	JPCN3en-ja	2019/07/27 12:53:49	3195	0.000000	0.000000	0.000000	-	-	-	-	-	-	-	NMT	Yes	Transformer Base (+ASPEC data), relative position, BT, r2l reranking, checkpoint ensemble
9	goku20	JPCN3en-ja	2020/09/21 12:28:46	4103	0.000000	0.000000	0.000000	-	-	-	-	-	-	-	NMT	No	mBART pre-training transformer, single model
10	goku20	JPCN3en-ja	2020/09/22 00:19:09	4116	0.000000	0.000000	0.000000	-	-	-	-	-	-	-	NMT	No	mBART pre-training transformer, ensemble of 3 models
11	tpt_wat	JPCN3en-ja	2021/04/27 02:29:11	5708	0.893443	0.893443	0.893443	-	-	-	-	-	-	-	NMT	No	Base Transformer model with joint vocab, size 8k
12	Bering Lab	JPCN3en-ja	2021/05/04 20:03:17	6375	0.896949	0.896949	0.896949	-	-	-	-	-	-	-	NMT	Yes	Transformer Ensemble with additional crawled parallel corpus
13	sakura	JPCN3en-ja	2024/08/08 19:53:13	7278	0.000000	0.000000	0.000000	-	-	-	-	-	-	-	NMT	No	LLM: Rakuten/RakutenAI-7B-chat Fine-Tuned with JPC Corpus in six direction (En-Ja, Ja-En, Ko-Ja, Ja-Ko, Zh-Ja, Ja-Zh) - Best
14	sakura	JPCN3en-ja	2024/08/09 00:52:14	7308	0.000000	0.000000	0.000000	-	-	-	-	-	-	-	NMT	No	LLM: Rakuten/RakutenAI-7B-chat Fine-Tuned with JPC Corpus in six direction (En-Ja, Ja-En, Ko-Ja, Ja-Ko, Zh-Ja, Ja-Zh)

Notice:

This table is sorted by the leftmost segmenters. You can change the segmenter used to sort by clicking each segmenter link.
Adequacy-Fluency Metrics (AMFM) is a two-dimensional automatic evaluation metric for machine translation, designed to operate at the sentence level. It is based on adequacy and fluency, to decouple semantic and syntactic components of the translation process to provide a balanced view on translation quality.
AMFM is calculated without tokenizers.
The detail of AMFM is shown on the following paper: "Adequacy–Fluency Metrics: Evaluating MT in the Continuous Space Model Framework" [pdf]. Invited Talk in WAT2015 also helps understanding [slide].

HUMAN (WAT2022)

Notice:

HUMAN (WAT2022) is the result of the Pairwise Crowdsourcing Evaluation on WAT2022.
HUMAN (WAT2022) was evaluated by 5 different workers and the final decision is made by the voting of the judgements.

HUMAN (WAT2021)

Notice:

HUMAN (WAT2021) is the result of the Pairwise Crowdsourcing Evaluation on WAT2021.
HUMAN (WAT2021) was evaluated by 5 different workers and the final decision is made by the voting of the judgements.

HUMAN (WAT2020)

Notice:

HUMAN (WAT2020) is the result of the Pairwise Crowdsourcing Evaluation on WAT2020.
HUMAN (WAT2020) was evaluated by 5 different workers and the final decision is made by the voting of the judgements.

HUMAN (WAT2019)

Notice:

HUMAN (WAT2019) is the result of the Pairwise Crowdsourcing Evaluation on WAT2019.
HUMAN (WAT2019) was evaluated by 5 different workers and the final decision is made by the voting of the judgements.

HUMAN (WAT2018)

Notice:

HUMAN (WAT2018) is the result of the Pairwise Crowdsourcing Evaluation on WAT2018.
HUMAN (WAT2018) was evaluated by 5 different workers and the final decision is made by the voting of the judgements.

HUMAN (WAT2017)

Notice:

HUMAN (WAT2017) is the result of the Pairwise Crowdsourcing Evaluation on WAT2017.
HUMAN (WAT2017) was evaluated by 5 different workers and the final decision is made by the voting of the judgements.

HUMAN (WAT2016)

Notice:

HUMAN (WAT2016) is the result of the Pairwise Crowdsourcing Evaluation on WAT2016.
HUMAN (WAT2016) was evaluated by 5 different workers and the final decision is made by the voting of the judgements.

HUMAN (WAT2015)

Notice:

HUMAN (WAT2015) is the result of the Pairwise Crowdsourcing Evaluation on WAT2015.
HUMAN (WAT2015) was evaluated by 5 different workers and the final decision is made by the voting of the judgements.
The detail of the evaluation can be found in the pdf document (PDF file).

HUMAN (WAT2014)

Notice:

HUMAN (WAT2014) is the result of the Pairwise Crowdsourcing Evaluation on WAT2014.
HUMAN (WAT2014) was evaluated by 3 different workers and the final decision is made by the voting of the judgements.
The detail of the evaluation can be found in the pdf document (PDF file).

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

WAT The Workshop on Asian Translation Evaluation Results

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

WAT

The Workshop on Asian Translation

Evaluation Results