opus-mt-tc-bible-big-bnt-deu_eng_fra_por_spa

Model Details
Uses
Risks, Limitations and Biases
How to Get Started With the Model
Training
Evaluation
Citation Information
Acknowledgements

Model Details

Neural machine translation model for translating from Bantu languages (bnt) to unknown (deu+eng+fra+por+spa).

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: 2024-05-30
License: Apache-2.0
Language(s):
- Source Language(s): bas bem bnt bss cce cjk cwe dig dug gog gwr hay heh her jmc kam kdc kdn kik kin kki kkj kmb kng kon ksb kua ldi lem lin lon lsm lua lug luy mcp myx nbl nde ndo nim nnb nso nuj nya nyf nyn nyo nyy old ozm pkb rim run seh sna sot ssw suk swa swc swh sxb thk tlj toh toi tsn tso tum umb ven vmw vun wmw xho xog zul
- Target Language(s): deu eng fra por spa
- Valid Target Language Labels: >>deu<< >>eng<< >>fra<< >>por<< >>spa<< >>xxx<<
Original Model: opusTCv20230926max50+bt+jhubc_transformer-big_2024-05-30.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. >>deu<<

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 = [
    ">>deu<< Replace this with text in an accepted source language.",
    ">>spa<< This is the second sentence."
]

model_name = "pytorch-models/opus-mt-tc-bible-big-bnt-deu_eng_fra_por_spa"
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) )

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-bible-big-bnt-deu_eng_fra_por_spa")
print(pipe(">>deu<< Replace this with text in an accepted source language."))

Training

Data: opusTCv20230926max50+bt+jhubc (source)
Pre-processing: SentencePiece (spm32k,spm32k)
Model Type: transformer-big
Original MarianNMT Model: opusTCv20230926max50+bt+jhubc_transformer-big_2024-05-30.zip
Training Scripts: GitHub Repo

Evaluation

Model scores at the OPUS-MT dashboard
test set translations: opusTCv20230926max50+bt+jhubc_transformer-big_2024-05-29.test.txt
test set scores: opusTCv20230926max50+bt+jhubc_transformer-big_2024-05-29.eval.txt
benchmark results: benchmark_results.txt
benchmark output: benchmark_translations.zip

langpair	testset	chr-F	BLEU	#sent	#words
run-deu	tatoeba-test-v2021-08-07	0.43836	26.1	1752	10562
run-eng	tatoeba-test-v2021-08-07	0.54089	39.4	1703	10041
run-fra	tatoeba-test-v2021-08-07	0.46240	26.1	1274	7479
run-spa	tatoeba-test-v2021-08-07	0.46496	25.8	963	5167
swa-eng	tatoeba-test-v2021-08-07	0.59947	45.9	387	2508
lin-eng	flores101-devtest	0.40858	16.9	1012	24721
nso-eng	flores101-devtest	0.49866	26.5	1012	24721
sna-fra	flores101-devtest	0.40134	14.3	1012	28343
swh-deu	flores101-devtest	0.43073	14.2	1012	25094
zul-fra	flores101-devtest	0.43723	17.4	1012	28343
zul-por	flores101-devtest	0.41886	15.9	1012	26519
bem-eng	flores200-devtest	0.42350	18.1	1012	24721
kin-eng	flores200-devtest	0.46183	21.9	1012	24721
kin-fra	flores200-devtest	0.40139	14.7	1012	28343
lin-eng	flores200-devtest	0.42073	18.1	1012	24721
nso-eng	flores200-devtest	0.51453	28.4	1012	24721
nso-fra	flores200-devtest	0.41065	16.1	1012	28343
nya-eng	flores200-devtest	0.44398	20.2	1012	24721
run-eng	flores200-devtest	0.42987	18.9	1012	24721
sna-eng	flores200-devtest	0.45917	21.1	1012	24721
sna-fra	flores200-devtest	0.41153	15.2	1012	28343
sot-eng	flores200-devtest	0.51854	26.9	1012	24721
sot-fra	flores200-devtest	0.41340	15.8	1012	28343
ssw-eng	flores200-devtest	0.44925	20.7	1012	24721
swh-deu	flores200-devtest	0.44937	15.6	1012	25094
swh-eng	flores200-devtest	0.60107	37.0	1012	24721
swh-fra	flores200-devtest	0.50257	23.5	1012	28343
swh-por	flores200-devtest	0.49475	22.8	1012	26519
swh-spa	flores200-devtest	0.42866	15.3	1012	29199
tsn-eng	flores200-devtest	0.45365	19.9	1012	24721
tso-eng	flores200-devtest	0.46882	22.8	1012	24721
xho-eng	flores200-devtest	0.52500	28.8	1012	24721
xho-fra	flores200-devtest	0.44642	18.7	1012	28343
xho-por	flores200-devtest	0.42517	16.8	1012	26519
zul-eng	flores200-devtest	0.53428	29.5	1012	24721
zul-fra	flores200-devtest	0.45383	19.0	1012	28343
zul-por	flores200-devtest	0.43537	17.4	1012	26519
bem-eng	ntrex128	0.43168	19.1	1997	47673
kin-eng	ntrex128	0.46996	20.8	1997	47673
kin-fra	ntrex128	0.40765	14.7	1997	53481
kin-spa	ntrex128	0.41552	15.9	1997	54107
nde-eng	ntrex128	0.42744	17.1	1997	47673
nso-eng	ntrex128	0.47231	21.5	1997	47673
nso-spa	ntrex128	0.40135	15.2	1997	54107
nya-eng	ntrex128	0.47072	23.3	1997	47673
nya-spa	ntrex128	0.41006	16.2	1997	54107
ssw-eng	ntrex128	0.48682	23.5	1997	47673
ssw-spa	ntrex128	0.40839	15.9	1997	54107
swa-deu	ntrex128	0.43880	14.1	1997	48761
swa-eng	ntrex128	0.58527	35.4	1997	47673
swa-fra	ntrex128	0.47344	19.7	1997	53481
swa-por	ntrex128	0.46292	19.1	1997	51631
swa-spa	ntrex128	0.48780	22.9	1997	54107
tsn-eng	ntrex128	0.50413	25.3	1997	47673
tsn-fra	ntrex128	0.41912	15.8	1997	53481
tsn-por	ntrex128	0.41090	15.3	1997	51631
tsn-spa	ntrex128	0.42979	17.7	1997	54107
ven-eng	ntrex128	0.43364	18.4	1997	47673
xho-eng	ntrex128	0.50778	26.5	1997	47673
xho-fra	ntrex128	0.41066	15.1	1997	53481
xho-spa	ntrex128	0.42129	16.7	1997	54107
zul-eng	ntrex128	0.50361	26.9	1997	47673
zul-fra	ntrex128	0.40779	15.0	1997	53481
zul-spa	ntrex128	0.41836	16.9	1997	54107
kin-eng	tico19-test	0.42280	18.8	2100	56323
lin-eng	tico19-test	0.41495	18.4	2100	56323
lug-eng	tico19-test	0.43948	22.2	2100	56323
swa-eng	tico19-test	0.58126	34.5	2100	56315
swa-fra	tico19-test	0.46470	20.5	2100	64661
swa-por	tico19-test	0.49374	22.8	2100	62729
swa-spa	tico19-test	0.50214	24.5	2100	66563
zul-eng	tico19-test	0.55678	32.2	2100	56804
zul-fra	tico19-test	0.43797	18.6	2100	64661
zul-por	tico19-test	0.45560	19.9	2100	62729
zul-spa	tico19-test	0.46505	21.4	2100	66563

Citation Information

Publications: Democratizing neural machine translation with OPUS-MT and 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.)

@article{tiedemann2023democratizing,
  title={Democratizing neural machine translation with {OPUS-MT}},
  author={Tiedemann, J{\"o}rg and Aulamo, Mikko and Bakshandaeva, Daria and Boggia, Michele and Gr{\"o}nroos, Stig-Arne and Nieminen, Tommi and Raganato, Alessandro and Scherrer, Yves and Vazquez, Raul and Virpioja, Sami},
  journal={Language Resources and Evaluation},
  number={58},
  pages={713--755},
  year={2023},
  publisher={Springer Nature},
  issn={1574-0218},
  doi={10.1007/s10579-023-09704-w}
}

@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 HPLT project, funded by the European Union’s Horizon Europe research and innovation programme under grant agreement No 101070350. We are also grateful for the generous computational resources and IT infrastructure provided by CSC -- IT Center for Science, Finland, and the EuroHPC supercomputer LUMI.

Model conversion info

transformers version: 4.45.1
OPUS-MT git hash: a0ea3b3
port time: Mon Oct 7 21:44:56 EEST 2024
port machine: LM0-400-22516.local

Helsinki-NLP
/

opus-mt-tc-bible-big-bnt-deu_eng_fra_por_spa