Helsinki-NLP/opus-mt-en-poz

eng-poz

• source group: English

• target group: Malayo-Polynesian languages

• OPUS readme: eng-poz

• model: transformer

• source language(s): eng

• target language(s): akl_Latn ceb cha dtp fij gil haw hil iba ilo ind jav jav_Java lkt mad mah max_Latn min mlg mri nau niu pag pau rap smo sun tah tet tmw_Latn ton tvl war zlm_Latn zsm_Latn

• 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)

• download original weights: opus-2020-07-27.zip

• test set translations: opus-2020-07-27.test.txt

• test set scores: opus-2020-07-27.eval.txt

Benchmarks

testset	BLEU	chr-F
Tatoeba-test.eng-akl.eng.akl	1.3	0.086
Tatoeba-test.eng-ceb.eng.ceb	10.2	0.426
Tatoeba-test.eng-cha.eng.cha	1.9	0.196
Tatoeba-test.eng-dtp.eng.dtp	0.4	0.121
Tatoeba-test.eng-fij.eng.fij	31.0	0.463
Tatoeba-test.eng-gil.eng.gil	45.4	0.635
Tatoeba-test.eng-haw.eng.haw	0.6	0.104
Tatoeba-test.eng-hil.eng.hil	14.4	0.498
Tatoeba-test.eng-iba.eng.iba	17.4	0.414
Tatoeba-test.eng-ilo.eng.ilo	33.1	0.585
Tatoeba-test.eng-jav.eng.jav	6.5	0.309
Tatoeba-test.eng-lkt.eng.lkt	0.5	0.065
Tatoeba-test.eng-mad.eng.mad	1.7	0.156
Tatoeba-test.eng-mah.eng.mah	12.7	0.391
Tatoeba-test.eng-mlg.eng.mlg	30.3	0.504
Tatoeba-test.eng-mri.eng.mri	8.2	0.316
Tatoeba-test.eng-msa.eng.msa	30.4	0.561
Tatoeba-test.eng.multi	16.2	0.410
Tatoeba-test.eng-nau.eng.nau	0.6	0.087
Tatoeba-test.eng-niu.eng.niu	33.2	0.482
Tatoeba-test.eng-pag.eng.pag	19.4	0.555
Tatoeba-test.eng-pau.eng.pau	1.0	0.124
Tatoeba-test.eng-rap.eng.rap	1.4	0.090
Tatoeba-test.eng-smo.eng.smo	12.9	0.407
Tatoeba-test.eng-sun.eng.sun	15.5	0.364
Tatoeba-test.eng-tah.eng.tah	9.5	0.295
Tatoeba-test.eng-tet.eng.tet	1.2	0.146
Tatoeba-test.eng-ton.eng.ton	23.7	0.484
Tatoeba-test.eng-tvl.eng.tvl	32.5	0.549
Tatoeba-test.eng-war.eng.war	12.6	0.432

System Info:

• hf_name: eng-poz

• source_languages: eng

• target_languages: poz

• opus_readme_url: https://github.com/Helsinki-NLP/Tatoeba-Challenge/tree/master/models/eng-poz/README.md

• original_repo: Tatoeba-Challenge

• tags: ['translation']

• languages: ['en', 'poz']

• src_constituents: {'eng'}

• tgt_constituents: set()

• src_multilingual: False

• tgt_multilingual: True

• prepro: normalization + SentencePiece (spm32k,spm32k)

• url_model: https://object.pouta.csc.fi/Tatoeba-MT-models/eng-poz/opus-2020-07-27.zip

• url_test_set: https://object.pouta.csc.fi/Tatoeba-MT-models/eng-poz/opus-2020-07-27.test.txt

• src_alpha3: eng

• tgt_alpha3: poz

• short_pair: en-poz

• chrF2_score: 0.41

• bleu: 16.2

• brevity_penalty: 1.0

• ref_len: 66803.0

• src_name: English

• tgt_name: Malayo-Polynesian languages

• train_date: 2020-07-27

• src_alpha2: en

• tgt_alpha2: poz

• prefer_old: False

• long_pair: eng-poz

• helsinki_git_sha: 480fcbe0ee1bf4774bcbe6226ad9f58e63f6c535

• transformers_git_sha: 2207e5d8cb224e954a7cba69fa4ac2309e9ff30b

• port_machine: brutasse

• port_time: 2020-08-21-14:41