opus-mt-tc-big-zle-zlw

Neural machine translation model for translating from East Slavic languages (zle) to West Slavic languages (zlw).

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: 2022-03-23
source language(s): bel rus ukr
target language(s): ces pol
valid target language labels: >>ces<< >>pol<<
model: transformer-big
data: opusTCv20210807+bt (source)
tokenization: SentencePiece (spm32k,spm32k)
original model: opusTCv20210807+bt_transformer-big_2022-03-23.zip
more information released models: OPUS-MT zle-zlw README
more information about the model: MarianMT

This is a multilingual translation model with multiple target languages. A sentence initial language token is required in the form of >>id<< (id = valid target language ID), e.g. >>ces<<

Usage

A short example code:

from transformers import MarianMTModel, MarianTokenizer

src_text = [
    ">>pol<< Это метафора.",
    ">>pol<< Что вы делали?"
]

model_name = "pytorch-models/opus-mt-tc-big-zle-zlw"
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:
#     To metafora.
#     Co robiliście?

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

from transformers import pipeline
pipe = pipeline("translation", model="Helsinki-NLP/opus-mt-tc-big-zle-zlw")
print(pipe(">>pol<< Это метафора."))

# expected output: To metafora.

Benchmarks

test set translations: opusTCv20210807+bt_transformer-big_2022-03-23.test.txt
test set scores: opusTCv20210807+bt_transformer-big_2022-03-23.eval.txt
benchmark results: benchmark_results.txt
benchmark output: benchmark_translations.zip

langpair	testset	chr-F	BLEU	#sent	#words
bel-pol	tatoeba-test-v2021-08-07	0.65517	47.1	287	1706
rus-ces	tatoeba-test-v2021-08-07	0.69695	53.4	2934	16831
rus-pol	tatoeba-test-v2021-08-07	0.72176	53.7	3543	21505
ukr-ces	tatoeba-test-v2021-08-07	0.73149	58.0	1787	8550
ukr-pol	tatoeba-test-v2021-08-07	0.74649	57.0	2519	13201
bel-ces	flores101-devtest	0.41248	11.1	1012	22101
bel-pol	flores101-devtest	0.42240	10.2	1012	22520
rus-ces	flores101-devtest	0.50971	23.1	1012	22101
rus-pol	flores101-devtest	0.48672	18.4	1012	22520
ukr-ces	flores101-devtest	0.52482	25.1	1012	22101
ukr-pol	flores101-devtest	0.48790	18.8	1012	22520
rus-ces	newstest2012	0.45834	18.8	3003	65456
rus-ces	newstest2013	0.52364	26.0	3000	57250

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: 1bdabf7
port time: Thu Mar 24 00:50:29 EET 2022
port machine: LM0-400-22516.local

Helsinki-NLP
/

opus-mt-tc-big-zle-zlw