language:
- be
- es
- ru
- rue
- uk
- zle
tags:
- translation
- opus-mt-tc
license: cc-by-4.0
model-index:
- name: opus-mt-tc-big-zle-es
results:
- task:
name: Translation rus-spa
type: translation
args: rus-spa
dataset:
name: flores101-devtest
type: flores_101
args: rus spa devtest
metrics:
- name: BLEU
type: bleu
value: 22.5
- task:
name: Translation ukr-spa
type: translation
args: ukr-spa
dataset:
name: flores101-devtest
type: flores_101
args: ukr spa devtest
metrics:
- name: BLEU
type: bleu
value: 22.7
- task:
name: Translation bel-spa
type: translation
args: bel-spa
dataset:
name: tatoeba-test-v2021-08-07
type: tatoeba_mt
args: bel-spa
metrics:
- name: BLEU
type: bleu
value: 46.3
- task:
name: Translation rus-spa
type: translation
args: rus-spa
dataset:
name: tatoeba-test-v2021-08-07
type: tatoeba_mt
args: rus-spa
metrics:
- name: BLEU
type: bleu
value: 52.3
- task:
name: Translation ukr-spa
type: translation
args: ukr-spa
dataset:
name: tatoeba-test-v2021-08-07
type: tatoeba_mt
args: ukr-spa
metrics:
- name: BLEU
type: bleu
value: 51.6
- task:
name: Translation rus-spa
type: translation
args: rus-spa
dataset:
name: newstest2012
type: wmt-2012-news
args: rus-spa
metrics:
- name: BLEU
type: bleu
value: 29
- task:
name: Translation rus-spa
type: translation
args: rus-spa
dataset:
name: newstest2013
type: wmt-2013-news
args: rus-spa
metrics:
- name: BLEU
type: bleu
value: 31.7
opus-mt-tc-big-zle-es
Neural machine translation model for translating from East Slavic languages (zle) to Spanish (es).
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): bel rue rus ukr
- target language(s): spa
- 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 zle-spa README
Usage
A short example code:
from transformers import MarianMTModel, MarianTokenizer
src_text = [
"Том був п'яничкою.",
"Он достаточно взрослый, чтобы путешествовать одному."
]
model_name = "pytorch-models/opus-mt-tc-big-zle-es"
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:
# Tom era un borracho.
# Es lo suficientemente mayor como para viajar solo.
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-zle-es")
print(pipe("Том був п'яничкою."))
# expected output: Tom era un borracho.
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 |
---|---|---|---|---|---|
bel-spa | tatoeba-test-v2021-08-07 | 0.65523 | 46.3 | 205 | 1412 |
rus-spa | tatoeba-test-v2021-08-07 | 0.69933 | 52.3 | 10506 | 75246 |
ukr-spa | tatoeba-test-v2021-08-07 | 0.68862 | 51.6 | 10115 | 59284 |
bel-spa | flores101-devtest | 0.44744 | 14.1 | 1012 | 29199 |
rus-spa | flores101-devtest | 0.50880 | 22.5 | 1012 | 29199 |
ukr-spa | flores101-devtest | 0.50943 | 22.7 | 1012 | 29199 |
rus-spa | newstest2012 | 0.55185 | 29.0 | 3003 | 79006 |
rus-spa | newstest2013 | 0.56826 | 31.7 | 3000 | 70528 |
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 00:12:49 EET 2022
- port machine: LM0-400-22516.local