Edit model card

opus-mt-tc-big-zlw-zle

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

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.

@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

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. >>bel<<

Usage

A short example code:

from transformers import MarianMTModel, MarianTokenizer

src_text = [
    ">>rus<< Je vystudovaný právník.",
    ">>rus<< Gdzie jest moja książka ?"
]

model_name = "pytorch-models/opus-mt-tc-big-zlw-zle"
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:
#     Он дипломированный юрист.
#     Где моя книга?

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-zlw-zle")
print(pipe(">>rus<< Je vystudovaný právník."))

# expected output: Он дипломированный юрист.

Benchmarks

langpair testset chr-F BLEU #sent #words
ces-rus tatoeba-test-v2021-08-07 0.73154 56.4 2934 17790
ces-ukr tatoeba-test-v2021-08-07 0.69934 53.0 1787 8891
pol-bel tatoeba-test-v2021-08-07 0.51039 29.4 287 1730
pol-rus tatoeba-test-v2021-08-07 0.73156 55.3 3543 22067
pol-ukr tatoeba-test-v2021-08-07 0.68247 48.6 2519 13535
ces-rus flores101-devtest 0.52316 24.2 1012 23295
ces-ukr flores101-devtest 0.52261 22.9 1012 22810
pol-rus flores101-devtest 0.49414 20.1 1012 23295
pol-ukr flores101-devtest 0.48250 18.3 1012 22810
ces-rus newstest2012 0.49469 21.0 3003 64790
ces-rus newstest2013 0.54197 27.2 3000 58560

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 04:13:23 EET 2022
  • port machine: LM0-400-22516.local
Downloads last month
1
Hosted inference API
This model can be loaded on the Inference API on-demand.

Evaluation results