Hugging Face's logo

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

Translation
Transformers
PyTorch
TensorFlow
Safetensors
marian
text2text-generation
opus-mt-tc
Eval Results
Inference Endpoints
Model card Files Community
2

opus-mt-tc-big-zls-itc

Table of Contents

Model Details

Neural machine translation model for translating from South Slavic languages (zls) 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: 2022-08-10
  • License: CC-BY-4.0
  • Language(s):
    • Source Language(s): bos_Latn bul hbs hrv mkd slv srp_Cyrl srp_Latn
    • Target Language(s): fra ita por ron spa
    • Language Pair(s): bul-fra bul-ita bul-por bul-ron bul-spa hbs-fra hbs-ita hbs-spa hrv-fra hrv-ita hrv-por hrv-ron hrv-spa mkd-fra mkd-ita mkd-por mkd-ron mkd-spa slv-fra slv-ita slv-por slv-ron slv-spa srp_Cyrl-fra srp_Cyrl-ita srp_Cyrl-por srp_Cyrl-ron srp_Cyrl-spa srp_Latn-ita
    • 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<< >>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<< >>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_2022-08-10.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. >>fra<<

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<< Dobar dan, kako si?",
    ">>spa<< Znam da je ovo čudno."
]

model_name = "pytorch-models/opus-mt-tc-big-zls-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:
#     Bonjour, comment allez-vous ?
#     Sé que esto es raro.

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-zls-itc")
print(pipe(">>fra<< Dobar dan, kako si?"))

# expected output: Bonjour, comment allez-vous ?

Training

Evaluation

langpair testset chr-F BLEU #sent #words
bul-fra tatoeba-test-v2021-08-07 0.68971 52.9 446 3669
bul-ita tatoeba-test-v2021-08-07 0.66412 45.1 2500 16951
bul-spa tatoeba-test-v2021-08-07 0.66672 49.7 286 1783
hbs-fra tatoeba-test-v2021-08-07 0.66434 48.1 474 3370
hbs-ita tatoeba-test-v2021-08-07 0.72381 53.5 534 3208
hbs-spa tatoeba-test-v2021-08-07 0.73105 58.0 607 3766
hrv-fra tatoeba-test-v2021-08-07 0.62800 44.3 258 1943
hrv-spa tatoeba-test-v2021-08-07 0.71370 57.5 254 1702
mkd-spa tatoeba-test-v2021-08-07 0.75366 62.1 217 1121
srp_Latn-ita tatoeba-test-v2021-08-07 0.76045 59.6 212 1292
bul-fra flores101-devtest 0.60640 34.4 1012 28343
bul-ita flores101-devtest 0.54135 24.0 1012 27306
bul-por flores101-devtest 0.59322 32.4 1012 26519
bul-ron flores101-devtest 0.55558 27.1 1012 26799
bul-spa flores101-devtest 0.50962 22.4 1012 29199
hrv-fra flores101-devtest 0.59349 33.1 1012 28343
hrv-ita flores101-devtest 0.52980 23.5 1012 27306
hrv-por flores101-devtest 0.57402 30.2 1012 26519
hrv-ron flores101-devtest 0.53650 25.9 1012 26799
hrv-spa flores101-devtest 0.50161 21.5 1012 29199
mkd-fra flores101-devtest 0.60801 35.2 1012 28343
mkd-ita flores101-devtest 0.53543 23.9 1012 27306
mkd-por flores101-devtest 0.59648 33.9 1012 26519
mkd-ron flores101-devtest 0.54998 28.0 1012 26799
mkd-spa flores101-devtest 0.51079 22.8 1012 29199
slv-fra flores101-devtest 0.58233 31.5 1012 28343
slv-ita flores101-devtest 0.52390 22.4 1012 27306
slv-por flores101-devtest 0.56436 29.0 1012 26519
slv-ron flores101-devtest 0.53116 25.0 1012 26799
slv-spa flores101-devtest 0.49621 21.1 1012 29199
srp_Cyrl-fra flores101-devtest 0.62110 36.0 1012 28343
srp_Cyrl-ita flores101-devtest 0.54083 23.9 1012 27306
srp_Cyrl-por flores101-devtest 0.61248 34.9 1012 26519
srp_Cyrl-ron flores101-devtest 0.56235 28.8 1012 26799
srp_Cyrl-spa flores101-devtest 0.51698 22.8 1012 29199

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: Fri Aug 12 23:59:29 EEST 2022
  • port machine: LM0-400-22516.local
Downloads last month
12
Safetensors
Model size
237M 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-zls-itc 7

Evaluation results

View on Papers With Code