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): dan fao isl nno nob nor swe
target language(s): eng
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 gmq-eng README

Usage

A short example code:

python

from transformers import MarianMTModel, MarianTokenizer

src_text = [
    "Han var synligt nervøs.",
    "Inte ens Tom själv var övertygad."
]

model_name = "pytorch-models/opus-mt-tc-big-gmq-en"
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:
#     He was visibly nervous.
#     Even Tom was not convinced.

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-gmq-en")
print(pipe("Han var synligt nervøs."))

# expected output: He was visibly nervous.

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
dan-eng	tatoeba-test-v2021-08-07	0.78292	65.9	10795	79684
fao-eng	tatoeba-test-v2021-08-07	0.47467	30.1	294	1984
isl-eng	tatoeba-test-v2021-08-07	0.68346	53.3	2503	19788
nno-eng	tatoeba-test-v2021-08-07	0.69788	56.1	460	3524
nob-eng	tatoeba-test-v2021-08-07	0.73524	60.2	4539	36823
swe-eng	tatoeba-test-v2021-08-07	0.77665	66.4	10362	68513
dan-eng	flores101-devtest	0.72322	49.3	1012	24721
isl-eng	flores101-devtest	0.59616	34.2	1012	24721
nob-eng	flores101-devtest	0.68224	44.2	1012	24721
swe-eng	flores101-devtest	0.72042	49.8	1012	24721
isl-eng	newsdev2021.is-en	0.56709	30.4	2004	46383
isl-eng	newstest2021.is-en	0.57756	34.4	1000	22529

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 19:13:11 EEST 2022
port machine: LM0-400-22516.local