opus-mt-tc-big-itc-he

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 Italic languages (itc) to Hebrew (he).

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): cat fra glg ita lad_Latn por ron spa
- Target Language(s): heb
- Language Pair(s): cat-heb fra-heb glg-heb ita-heb por-heb ron-heb spa-heb
- Valid Target Language Labels:
Original Model: opusTCv20210807_transformer-big_2022-08-03.zip
Resources for more information:
- OPUS-MT-train GitHub Repo
- More information about released models for this language pair: OPUS-MT itc-heb README
- More information about MarianNMT models in the transformers library
- [Tatoeba Translation Challenge](https://github.com/Helsinki-NLP/Tatoeba-Challenge/

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 = [
    "La María és feminista.",
    "Contribuyan en Tatoeba."
]

model_name = "pytorch-models/opus-mt-tc-big-itc-he"
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:
#     מרי היא פמיניסטית.
#     תרום לטאטואבה.

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-itc-he")
print(pipe("La María és feminista."))

# expected output: מרי היא פמיניסטית.

Training

Data: opusTCv20210807 (source)
Pre-processing: SentencePiece (spm32k,spm32k)
Model Type: transformer-big
Original MarianNMT Model: opusTCv20210807_transformer-big_2022-08-03.zip
Training Scripts: GitHub Repo

Evaluation

test set translations: opusTCv20210807_transformer-big_2022-08-03.test.txt
test set scores: opusTCv20210807_transformer-big_2022-08-03.eval.txt
benchmark results: benchmark_results.txt
benchmark output: benchmark_translations.zip

langpair	testset	chr-F	BLEU	#sent	#words
fra-heb	tatoeba-test-v2021-08-07	0.60539	39.6	3281	20655
ita-heb	tatoeba-test-v2021-08-07	0.60264	40.0	1706	9796
por-heb	tatoeba-test-v2021-08-07	0.63087	44.4	719	4423
spa-heb	tatoeba-test-v2021-08-07	0.63883	44.5	1849	12112
cat-heb	flores101-devtest	0.52457	23.0	1012	20749
fra-heb	flores101-devtest	0.52953	23.2	1012	20749
glg-heb	flores101-devtest	0.50918	20.8	1012	20749
ita-heb	flores101-devtest	0.49007	18.3	1012	20749
por-heb	flores101-devtest	0.53906	24.4	1012	20749
ron-heb	flores101-devtest	0.52103	22.1	1012	20749
spa-heb	flores101-devtest	0.47646	16.5	1012	20749

Citation Information

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",
}

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 00:02:03 EEST 2022
port machine: LM0-400-22516.local

Helsinki-NLP
/

opus-mt-tc-big-itc-he