dra-eng

opus-2020-06-28.zip

• dataset: opus
• model: transformer
• source language(s): kan mal tam tel
• target language(s): eng
• model: transformer
• pre-processing: normalization + SentencePiece (spm32k,spm32k)
• download: opus-2020-06-28.zip
• test set translations: opus-2020-06-28.test.txt
• test set scores: opus-2020-06-28.eval.txt

Benchmarks

testset	BLEU	chr-F
Tatoeba-test.kan-eng.kan.eng	13.5	0.361
Tatoeba-test.mal-eng.mal.eng	42.5	0.599
Tatoeba-test.multi.eng	31.5	0.510
Tatoeba-test.tam-eng.tam.eng	30.8	0.478
Tatoeba-test.tel-eng.tel.eng	17.1	0.378

opus-2020-07-26.zip

• dataset: opus
• model: transformer
• source language(s): kan mal tam tel
• target language(s): eng
• model: transformer
• pre-processing: normalization + SentencePiece (spm32k,spm32k)
• download: opus-2020-07-26.zip
• test set translations: opus-2020-07-26.test.txt
• test set scores: opus-2020-07-26.eval.txt

Benchmarks

testset	BLEU	chr-F
Tatoeba-test.kan-eng.kan.eng	8.6	0.288
Tatoeba-test.mal-eng.mal.eng	36.8	0.533
Tatoeba-test.multi.eng	25.5	0.443
Tatoeba-test.tam-eng.tam.eng	22.9	0.405
Tatoeba-test.tel-eng.tel.eng	12.2	0.322

opus2m-2020-08-12.zip

• dataset: opus2m
• model: transformer
• source language(s): kan mal tam tel
• target language(s): eng
• model: transformer
• pre-processing: normalization + SentencePiece (spm32k,spm32k)
• download: opus2m-2020-08-12.zip
• test set translations: opus2m-2020-08-12.test.txt
• test set scores: opus2m-2020-08-12.eval.txt

Benchmarks

testset	BLEU	chr-F
Tatoeba-test.kan-eng.kan.eng	9.1	0.312
Tatoeba-test.mal-eng.mal.eng	42.0	0.584
Tatoeba-test.multi.eng	30.0	0.493
Tatoeba-test.tam-eng.tam.eng	30.2	0.467
Tatoeba-test.tel-eng.tel.eng	15.9	0.378

opus1m+bt-2021-05-01.zip

• dataset: opus1m+bt
• model: transformer-align
• source language(s): kan mal tam tel
• target language(s): eng
• model: transformer-align
• pre-processing: normalization + SentencePiece (spm32k,spm32k)
• download: opus1m+bt-2021-05-01.zip
• test set translations: opus1m+bt-2021-05-01.test.txt
• test set scores: opus1m+bt-2021-05-01.eval.txt

Benchmarks

testset	BLEU	chr-F	#sent	#words	BP
Tatoeba-test.kan-eng	17.8	0.404	167	1252	1.000
Tatoeba-test.mal-eng	42.8	0.602	802	5558	0.985
Tatoeba-test.multi-eng	32.2	0.519	1541	10641	1.000
Tatoeba-test.tam-eng	27.4	0.460	311	2125	1.000
Tatoeba-test.tel-eng	19.5	0.405	261	1706	1.000
tico19-test.tam-eng	13.6	0.380	2100	56848	1.000

opus4m+btTCv20210807-2021-09-30.zip

• dataset: opus4m+btTCv20210807
• model: transformer
• source language(s): kan mal tam tcy tel
• target language(s): eng
• 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:
• download: opus4m+btTCv20210807-2021-09-30.zip
• test set translations: opus4m+btTCv20210807-2021-09-30.test.txt
• test set scores: opus4m+btTCv20210807-2021-09-30.eval.txt

Benchmarks

testset	BLEU	chr-F	#sent	#words	BP
Tatoeba-test-v2021-08-07.multi-eng	34.8	0.529	1541	10641	1.000
Tatoeba-test-v2021-08-07.multi-multi	34.8	0.529	1541	10641	1.000
tico19-test.tam-eng	22.9	0.505	2100	56848	1.000