This model is part of the OPUS-MT project, an effort to make neural machine translation models widely available and accessible for many languages in the world. All models are originally trained using the amazing framework of Marian NMT, an efficient NMT implementation written in pure C++. The models have been converted to pyTorch using the transformers library by huggingface. Training data is taken from OPUS and training pipelines use the procedures of OPUS-MT-train.

Publications: OPUS-MT – Building open translation services for the World and The Tatoeba Translation Challenge – Realistic Data Sets for Low Resource and Multilingual MT (Please, cite if you use this model.)

@inproceedings{tiedemann-thottingal-2020-opus,
    title = "{OPUS}-{MT} {--} Building open translation services for the World",
    author = {Tiedemann, J{\"o}rg  and Thottingal, Santhosh},
    booktitle = "Proceedings of the 22nd Annual Conference of the European Association for Machine Translation",
    month = nov,
    year = "2020",
    address = "Lisboa, Portugal",
    publisher = "European Association for Machine Translation",
    url = "https://aclanthology.org/2020.eamt-1.61",
    pages = "479--480",
}

@inproceedings{tiedemann-2020-tatoeba,
    title = "The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}",
    author = {Tiedemann, J{\"o}rg},
    booktitle = "Proceedings of the Fifth Conference on Machine Translation",
    month = nov,
    year = "2020",
    address = "Online",
    publisher = "Association for Computational Linguistics",
    url = "https://aclanthology.org/2020.wmt-1.139",
    pages = "1174--1182",
}

Model info

Release: 2022-03-09
source language(s): eng
target language(s): fra
model: transformer-big
data: opusTCv20210807+bt (source)
tokenization: SentencePiece (spm32k,spm32k)
original model: opusTCv20210807+bt_transformer-big_2022-03-09.zip
more information released models: OPUS-MT eng-fra README

Usage

A short example code:

python

from transformers import MarianMTModel, MarianTokenizer

src_text = [
    "The Portuguese teacher is very demanding.",
    "When was your last hearing test?"
]

model_name = "pytorch-models/opus-mt-tc-big-en-fr"
tokenizer = MarianTokenizer.from_pretrained(model_name)
model = MarianMTModel.from_pretrained(model_name)
translated = model.generate(**tokenizer(src_text, return_tensors="pt", padding=True))

for t in translated:
    print( tokenizer.decode(t, skip_special_tokens=True) )

# expected output:
#     Le professeur de portugais est très exigeant.
#     Quand a eu lieu votre dernier test auditif ?

You can also use OPUS-MT models with the transformers pipelines, for example:

python

from transformers import pipeline
pipe = pipeline("translation", model="Helsinki-NLP/opus-mt-tc-big-en-fr")
print(pipe("The Portuguese teacher is very demanding."))

# expected output: Le professeur de portugais est très exigeant.

Benchmarks

test set translations: opusTCv20210807+bt_transformer-big_2022-03-09.test.txt
test set scores: opusTCv20210807+bt_transformer-big_2022-03-09.eval.txt
benchmark results: benchmark_results.txt
benchmark output: benchmark_translations.zip

langpair	testset	chr-F	BLEU	#sent	#words
eng-fra	tatoeba-test-v2021-08-07	0.69621	53.2	12681	106378
eng-fra	flores101-devtest	0.72494	52.2	1012	28343
eng-fra	multi30k_test_2016_flickr	0.72361	52.4	1000	13505
eng-fra	multi30k_test_2017_flickr	0.72826	52.8	1000	12118
eng-fra	multi30k_test_2017_mscoco	0.73547	54.7	461	5484
eng-fra	multi30k_test_2018_flickr	0.66723	43.7	1071	15867
eng-fra	newsdiscussdev2015	0.60471	33.4	1500	27940
eng-fra	newsdiscusstest2015	0.64915	40.3	1500	27975
eng-fra	newssyscomb2009	0.58903	30.7	502	12331
eng-fra	news-test2008	0.55516	27.6	2051	52685
eng-fra	newstest2009	0.57907	30.0	2525	69263
eng-fra	newstest2010	0.60156	33.5	2489	66022
eng-fra	newstest2011	0.61632	35.0	3003	80626
eng-fra	newstest2012	0.59736	32.8	3003	78011
eng-fra	newstest2013	0.59700	34.6	3000	70037
eng-fra	newstest2014	0.66686	41.9	3003	77306
eng-fra	tico19-test	0.63022	40.6	2100	64661

Acknowledgements

The work is supported by the European Language Grid as pilot project 2866, by the FoTran project, funded by the European Research Council (ERC) under the European Union’s Horizon 2020 research and innovation programme (grant agreement No 771113), and the MeMAD project, funded by the European Union’s Horizon 2020 Research and Innovation Programme under grant agreement No 780069. We are also grateful for the generous computational resources and IT infrastructure provided by CSC -- IT Center for Science, Finland.

Model conversion info

transformers version: 4.16.2
OPUS-MT git hash: 3405783
port time: Wed Apr 13 17:07:05 EEST 2022
port machine: LM0-400-22516.local