mkh-mkh

opus-2021-02-16.zip

• dataset: opus
• model: transformer
• source language(s): eng kha khm mnw vie
• target language(s): eng kha khm mnw vie
• model: transformer
• pre-processing: normalization + SentencePiece (spm32k,spm32k)
• a sentence initial language token is required in the form of >>id<< (id = valid target language ID)
• valid language labels: >>eng<< >>vie<< >>khm<< >>khm_Latn<<
• download: opus-2021-02-16.zip
• test set translations: opus-2021-02-16.test.txt
• test set scores: opus-2021-02-16.eval.txt

Benchmarks

testset	BLEU	chr-F	#sent	#words	BP
Tatoeba-test.multi-multi	21.7	0.366	9194	68857	1.000

opus-2021-02-18.zip

• dataset: opus
• model: transformer
• source language(s): kha khm vie
• target language(s): kha khm vie
• model: transformer
• pre-processing: normalization + SentencePiece (spm32k,spm32k)
• a sentence initial language token is required in the form of >>id<< (id = valid target language ID)
• valid language labels: >>eng<< >>vie<< >>khm<< >>khm_Latn<<
• download: opus-2021-02-18.zip
• test set translations: opus-2021-02-18.test.txt
• test set scores: opus-2021-02-18.eval.txt

Benchmarks

testset	BLEU	chr-F	#sent	#words	BP
Tatoeba-test.kha-vie	2.1	0.104	4	39	0.920
Tatoeba-test.khm-vie	9.2	0.311	18	101	1.000
Tatoeba-test.multi-multi	21.7	0.366	9194	68857	1.000
Tatoeba-test.vie-kha	2.1	0.078	4	37	1.000
Tatoeba-test.vie-khm	1.4	0.151	18	37	1.000
tico19-test.eng-khm	1.4	0.290	2100	20941	1.000