Hugging Face's logo

Helsinki-NLP
/
opus-mt-tc-big-zle-itc

Translation
Transformers
PyTorch
TensorFlow
Safetensors
marian
text2text-generation
opus-mt-tc
Eval Results
Inference Endpoints
Model card Files Community
1
Edit model card

opus-mt-tc-big-zle-itc

Table of Contents

Model Details

Neural machine translation model for translating from East Slavic languages (zle) to Italic languages (itc).

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. Model Description:

  • Developed by: Language Technology Research Group at the University of Helsinki
  • Model Type: Translation (transformer-big)
  • Release: 2022-08-03
  • License: CC-BY-4.0
  • Language(s):
    • Source Language(s): bel rue rus ukr
    • Target Language(s): cat fra glg ita lad_Latn por ron spa
    • Language Pair(s): bel-fra bel-ita bel-spa rus-cat rus-fra rus-glg rus-ita rus-por rus-ron rus-spa ukr-cat ukr-fra ukr-glg ukr-ita ukr-por ukr-ron ukr-spa
    • Valid Target Language Labels: >>acf<< >>aoa<< >>arg<< >>ast<< >>cat<< >>cbk<< >>ccd<< >>cks<< >>cos<< >>cri<< >>crs<< >>dlm<< >>drc<< >>egl<< >>ext<< >>fab<< >>fax<< >>fra<< >>frc<< >>frm<< >>fro<< >>frp<< >>fur<< >>gcf<< >>gcf_Latn<< >>gcr<< >>glg<< >>hat<< >>idb<< >>ist<< >>ita<< >>itk<< >>kea<< >>kmv<< >>lad<< >>lad_Latn<< >>lat<< >>lat_Latn<< >>lij<< >>lld<< >>lmo<< >>lou<< >>mcm<< >>mfe<< >>mol<< >>mwl<< >>mxi<< >>mzs<< >>nap<< >>nrf<< >>oci<< >>osc<< >>osp<< >>osp_Latn<< >>pap<< >>pcd<< >>pln<< >>pms<< >>pob<< >>por<< >>pov<< >>pre<< >>pro<< >>qbb<< >>qhr<< >>rcf<< >>rgn<< >>roh<< >>ron<< >>ruo<< >>rup<< >>ruq<< >>scf<< >>scn<< >>sdc<< >>sdn<< >>spa<< >>spq<< >>spx<< >>src<< >>srd<< >>sro<< >>tmg<< >>tvy<< >>vec<< >>vkp<< >>wln<< >>xfa<< >>xum<<
  • Original Model: opusTCv20210807_transformer-big_2022-08-03.zip
  • Resources for more information:

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. >>cat<<

Uses

This model can be used for translation and text-to-text generation.

Risks, Limitations and Biases

CONTENT WARNING: Readers should be aware that the model is trained on various public data sets that may contain content that is disturbing, offensive, and can propagate historical and current stereotypes.

Significant research has explored bias and fairness issues with language models (see, e.g., Sheng et al. (2021) and Bender et al. (2021)).

How to Get Started With the Model

A short example code:

from transformers import MarianMTModel, MarianTokenizer

src_text = [
    ">>fra<< Вони не мої справжні батьки.",
    ">>por<< Мне нужно в школу."
]

model_name = "pytorch-models/opus-mt-tc-big-zle-itc"
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:
#     Ce ne sont pas mes vrais parents.
#     Tenho de ir para a escola.

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-itc")
print(pipe(">>fra<< Вони не мої справжні батьки."))

# expected output: Ce ne sont pas mes vrais parents.

Training

Evaluation

langpair testset chr-F BLEU #sent #words
bel-fra tatoeba-test-v2021-08-07 0.66784 49.1 283 2005
bel-ita tatoeba-test-v2021-08-07 0.64145 47.6 264 1681
bel-spa tatoeba-test-v2021-08-07 0.65485 46.9 205 1412
rus-fra tatoeba-test-v2021-08-07 0.68174 52.1 11490 80579
rus-ita tatoeba-test-v2021-08-07 0.63277 42.7 10045 71584
rus-por tatoeba-test-v2021-08-07 0.63606 42.6 10000 74713
rus-ron tatoeba-test-v2021-08-07 0.60796 37.5 782 4772
rus-spa tatoeba-test-v2021-08-07 0.69108 51.3 10506 75246
ukr-cat tatoeba-test-v2021-08-07 0.69275 52.9 456 2675
ukr-fra tatoeba-test-v2021-08-07 0.67392 51.3 10035 63227
ukr-ita tatoeba-test-v2021-08-07 0.69157 49.6 5000 27846
ukr-por tatoeba-test-v2021-08-07 0.64722 45.0 3372 21315
ukr-spa tatoeba-test-v2021-08-07 0.68409 50.7 10115 59284
bel-ast flores101-devtest 0.40942 8.7 1012 24572
bel-cat flores101-devtest 0.48374 16.8 1012 27304
bel-fra flores101-devtest 0.51278 19.4 1012 28343
bel-glg flores101-devtest 0.45665 15.3 1012 26582
bel-ita flores101-devtest 0.47204 14.6 1012 27306
bel-por flores101-devtest 0.49561 17.3 1012 26519
bel-ron flores101-devtest 0.46315 14.9 1012 26799
bel-spa flores101-devtest 0.46011 15.3 1012 29199
rus-ast flores101-devtest 0.45411 13.6 1012 24572
rus-cat flores101-devtest 0.55262 28.3 1012 27304
rus-fra flores101-devtest 0.59498 32.9 1012 28343
rus-glg flores101-devtest 0.51668 23.5 1012 26582
rus-ita flores101-devtest 0.52402 22.7 1012 27306
rus-oci flores101-devtest 0.42301 12.9 1012 27305
rus-por flores101-devtest 0.58045 31.4 1012 26519
rus-ron flores101-devtest 0.52560 24.7 1012 26799
rus-spa flores101-devtest 0.50622 21.8 1012 29199
ukr-ast flores101-devtest 0.45629 14.1 1012 24572
ukr-cat flores101-devtest 0.56383 29.5 1012 27304
ukr-fra flores101-devtest 0.60596 34.5 1012 28343
ukr-glg flores101-devtest 0.52217 24.2 1012 26582
ukr-ita flores101-devtest 0.52610 23.0 1012 27306
ukr-oci flores101-devtest 0.42937 13.7 1012 27305
ukr-por flores101-devtest 0.59036 32.5 1012 26519
ukr-ron flores101-devtest 0.53883 26.0 1012 26799
ukr-spa flores101-devtest 0.51018 22.5 1012 29199
rus-fra newstest2012 0.53481 25.0 3003 78011
rus-spa newstest2012 0.54814 28.7 3003 79006
rus-fra newstest2013 0.55745 29.0 3000 70037
rus-spa newstest2013 0.56582 31.5 3000 70528

Citation Information 

@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",
}

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: 8b9f0b0
  • port time: Sat Aug 13 00:01:33 EEST 2022
  • port machine: LM0-400-22516.local
Downloads last month
1
Safetensors
Model size
239M params
Tensor type
FP16
·
Inference Examples
Translation
This model does not have enough activity to be deployed to Inference API (serverless) yet. Increase its social visibility and check back later, or deploy to Inference Endpoints (dedicated) instead.

Spaces using Helsinki-NLP/opus-mt-tc-big-zle-itc 7

Evaluation results

View on Papers With Code