README.md

---
library_name: transformers
language:
- chm
- de
- en
- et
- fi
- fkv
- hu
- izh
- krl
- kv
- liv
- mdf
- mrj
- myv
- nl
- se
- sma
- smn
- udm
- vot

tags:
- translation
- opus-mt-tc-bible

license: apache-2.0
model-index:
- name: opus-mt-tc-bible-big-urj-deu_eng_nld
  results:
  - task:
      name: Translation multi-multi
      type: translation
      args: multi-multi
    dataset:
      name: tatoeba-test-v2020-07-28-v2023-09-26
      type: tatoeba_mt
      args: multi-multi
    metrics:
       - name: BLEU
         type: bleu
         value: 46.1
       - name: chr-F
         type: chrf
         value: 0.65088
---
# opus-mt-tc-bible-big-urj-deu_eng_nld

## Table of Contents
- [Model Details](#model-details)
- [Uses](#uses)
- [Risks, Limitations and Biases](#risks-limitations-and-biases)
- [How to Get Started With the Model](#how-to-get-started-with-the-model)
- [Training](#training)
- [Evaluation](#evaluation)
- [Citation Information](#citation-information)
- [Acknowledgements](#acknowledgements)

## Model Details

Neural machine translation model for translating from Uralic languages (urj) to unknown (deu+eng+nld).

This model is part of the [OPUS-MT project](https://github.com/Helsinki-NLP/Opus-MT), 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](https://marian-nmt.github.io/), 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](https://opus.nlpl.eu/) and training pipelines use the procedures of [OPUS-MT-train](https://github.com/Helsinki-NLP/Opus-MT-train).
**Model Description:**
- **Developed by:** Language Technology Research Group at the University of Helsinki
- **Model Type:** Translation (transformer-big)
- **Release**: 2024-08-18
- **License:** Apache-2.0
- **Language(s):**  
  - Source Language(s): chm est fin fkv hun izh koi kom kpv krl liv mdf mrj myv sma sme smn udm vot vro
  - Target Language(s): deu eng nld
  - Valid Target Language Labels: >>deu<< >>eng<< >>nld<< >>xxx<<
- **Original Model**: [opusTCv20230926max50+bt+jhubc_transformer-big_2024-08-18.zip](https://object.pouta.csc.fi/Tatoeba-MT-models/urj-deu+eng+nld/opusTCv20230926max50+bt+jhubc_transformer-big_2024-08-18.zip)
- **Resources for more information:**
  -  [OPUS-MT dashboard](https://opus.nlpl.eu/dashboard/index.php?pkg=opusmt&test=all&scoreslang=all&chart=standard&model=Tatoeba-MT-models/urj-deu%2Beng%2Bnld/opusTCv20230926max50%2Bbt%2Bjhubc_transformer-big_2024-08-18)
  - [OPUS-MT-train GitHub Repo](https://github.com/Helsinki-NLP/OPUS-MT-train)
  - [More information about MarianNMT models in the transformers library](https://huggingface.co/docs/transformers/model_doc/marian)
  - [Tatoeba Translation Challenge](https://github.com/Helsinki-NLP/Tatoeba-Challenge/)
  - [HPLT bilingual data v1 (as part of the Tatoeba Translation Challenge dataset)](https://hplt-project.org/datasets/v1)
  - [A massively parallel Bible corpus](https://aclanthology.org/L14-1215/)

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)](https://aclanthology.org/2021.acl-long.330.pdf) and [Bender et al. (2021)](https://dl.acm.org/doi/pdf/10.1145/3442188.3445922)).

## How to Get Started With the Model

A short example code:

```python
from transformers import MarianMTModel, MarianTokenizer

src_text = [
    ">>deu<< Jobb meghalni, mint úgy élni.",
    ">>eng<< Az algák miatt ilyen színű a tó."
]

model_name = "pytorch-models/opus-mt-tc-bible-big-urj-deu_eng_nld"
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:
#     Es ist besser zu sterben, als so zu leben.
#     Because of the algae, the lake is such a color.
```

You can also use OPUS-MT models with the transformers pipelines, for example:

```python
from transformers import pipeline
pipe = pipeline("translation", model="Helsinki-NLP/opus-mt-tc-bible-big-urj-deu_eng_nld")
print(pipe(">>deu<< Jobb meghalni, mint úgy élni."))

# expected output: Es ist besser zu sterben, als so zu leben.
```

## Training

- **Data**: opusTCv20230926max50+bt+jhubc ([source](https://github.com/Helsinki-NLP/Tatoeba-Challenge))
- **Pre-processing**: SentencePiece (spm32k,spm32k)
- **Model Type:**  transformer-big
- **Original MarianNMT Model**: [opusTCv20230926max50+bt+jhubc_transformer-big_2024-08-18.zip](https://object.pouta.csc.fi/Tatoeba-MT-models/urj-deu+eng+nld/opusTCv20230926max50+bt+jhubc_transformer-big_2024-08-18.zip)
- **Training Scripts**: [GitHub Repo](https://github.com/Helsinki-NLP/OPUS-MT-train)

## Evaluation

* [Model scores at the OPUS-MT dashboard](https://opus.nlpl.eu/dashboard/index.php?pkg=opusmt&test=all&scoreslang=all&chart=standard&model=Tatoeba-MT-models/urj-deu%2Beng%2Bnld/opusTCv20230926max50%2Bbt%2Bjhubc_transformer-big_2024-08-18)
* test set translations: [opusTCv20230926max50+bt+jhubc_transformer-big_2024-08-18.test.txt](https://object.pouta.csc.fi/Tatoeba-MT-models/urj-deu+eng+nld/opusTCv20230926max50+bt+jhubc_transformer-big_2024-08-18.test.txt)
* test set scores: [opusTCv20230926max50+bt+jhubc_transformer-big_2024-08-18.eval.txt](https://object.pouta.csc.fi/Tatoeba-MT-models/urj-deu+eng+nld/opusTCv20230926max50+bt+jhubc_transformer-big_2024-08-18.eval.txt)
* benchmark results: [benchmark_results.txt](benchmark_results.txt)
* benchmark output: [benchmark_translations.zip](benchmark_translations.zip)

| langpair | testset | chr-F | BLEU  | #sent | #words |
|----------|---------|-------|-------|-------|--------|
| multi-multi | tatoeba-test-v2020-07-28-v2023-09-26 | 0.65088 | 46.1 | 10000 | 78967 |

## Citation Information

* Publications: [Democratizing neural machine translation with OPUS-MT](https://doi.org/10.1007/s10579-023-09704-w) and [OPUS-MT – Building open translation services for the World](https://aclanthology.org/2020.eamt-1.61/) and [The Tatoeba Translation Challenge – Realistic Data Sets for Low Resource and Multilingual MT](https://aclanthology.org/2020.wmt-1.139/) (Please, cite if you use this model.)

```bibtex
@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](https://hplt-project.org/), 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](https://www.csc.fi/), Finland, and the [EuroHPC supercomputer LUMI](https://www.lumi-supercomputer.eu/).

## Model conversion info

* transformers version: 4.45.1
* OPUS-MT git hash: 0882077
* port time: Wed Oct  9 00:49:52 EEST 2024
* port machine: LM0-400-22516.local