language:
- be
- es
- ru
- uk
- zle
tags:
- translation
- opus-mt-tc
license: cc-by-4.0
model-index:
- name: opus-mt-tc-big-es-zle
results:
- task:
name: Translation spa-rus
type: translation
args: spa-rus
dataset:
name: flores101-devtest
type: flores_101
args: spa rus devtest
metrics:
- name: BLEU
type: bleu
value: 20.2
- task:
name: Translation spa-bel
type: translation
args: spa-bel
dataset:
name: tatoeba-test-v2021-08-07
type: tatoeba_mt
args: spa-bel
metrics:
- name: BLEU
type: bleu
value: 27.5
- task:
name: Translation spa-rus
type: translation
args: spa-rus
dataset:
name: tatoeba-test-v2021-08-07
type: tatoeba_mt
args: spa-rus
metrics:
- name: BLEU
type: bleu
value: 49
- task:
name: Translation spa-ukr
type: translation
args: spa-ukr
dataset:
name: tatoeba-test-v2021-08-07
type: tatoeba_mt
args: spa-ukr
metrics:
- name: BLEU
type: bleu
value: 42.3
- task:
name: Translation spa-rus
type: translation
args: spa-rus
dataset:
name: newstest2012
type: wmt-2012-news
args: spa-rus
metrics:
- name: BLEU
type: bleu
value: 24.6
- task:
name: Translation spa-rus
type: translation
args: spa-rus
dataset:
name: newstest2013
type: wmt-2013-news
args: spa-rus
metrics:
- name: BLEU
type: bleu
value: 26.9
opus-mt-tc-big-es-zle
Neural machine translation model for translating from Spanish (es) to East Slavic languages (zle).
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.
- 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",
}
Model info
- Release: 2022-03-23
- source language(s): spa
- target language(s): bel rus ukr
- valid target language labels: >>bel<< >>rus<< >>ukr<<
- model: transformer-big
- data: opusTCv20210807 (source)
- tokenization: SentencePiece (spm32k,spm32k)
- original model: opusTCv20210807_transformer-big_2022-03-23.zip
- more information released models: OPUS-MT spa-zle README
- more information about the model: MarianMT
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. >>bel<<
Usage
A short example code:
from transformers import MarianMTModel, MarianTokenizer
src_text = [
">>rus<< Su novela se vendió bien.",
">>ukr<< Quiero ir a Corea del Norte."
]
model_name = "pytorch-models/opus-mt-tc-big-es-zle"
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:
# Его роман хорошо продавался.
# Я хочу поїхати до Північної Кореї.
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-es-zle")
print(pipe(">>rus<< Su novela se vendió bien."))
# expected output: Его роман хорошо продавался.
Benchmarks
- test set translations: opusTCv20210807_transformer-big_2022-03-23.test.txt
- test set scores: opusTCv20210807_transformer-big_2022-03-23.eval.txt
- benchmark results: benchmark_results.txt
- benchmark output: benchmark_translations.zip
langpair | testset | chr-F | BLEU | #sent | #words |
---|---|---|---|---|---|
spa-bel | tatoeba-test-v2021-08-07 | 0.54506 | 27.5 | 205 | 1259 |
spa-rus | tatoeba-test-v2021-08-07 | 0.68523 | 49.0 | 10506 | 69242 |
spa-ukr | tatoeba-test-v2021-08-07 | 0.63502 | 42.3 | 10115 | 54544 |
spa-rus | flores101-devtest | 0.49913 | 20.2 | 1012 | 23295 |
spa-ukr | flores101-devtest | 0.47772 | 17.4 | 1012 | 22810 |
spa-rus | newstest2012 | 0.52436 | 24.6 | 3003 | 64790 |
spa-rus | newstest2013 | 0.54249 | 26.9 | 3000 | 58560 |
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: 1bdabf7
- port time: Thu Mar 24 03:35:13 EET 2022
- port machine: LM0-400-22516.local