tiedeman's picture
Initial commit
85107cb
metadata
language:
  - ca
  - es
  - fr
  - gl
  - it
  - lt
  - lv
  - pt
tags:
  - translation
  - opus-mt-tc
license: cc-by-4.0
model-index:
  - name: opus-mt-tc-big-itc-bat
    results:
      - task:
          name: Translation cat-lav
          type: translation
          args: cat-lav
        dataset:
          name: flores101-devtest
          type: flores_101
          args: cat lav devtest
        metrics:
          - name: BLEU
            type: bleu
            value: 21.9
          - name: chr-F
            type: chrf
            value: 0.52215
      - task:
          name: Translation cat-lit
          type: translation
          args: cat-lit
        dataset:
          name: flores101-devtest
          type: flores_101
          args: cat lit devtest
        metrics:
          - name: BLEU
            type: bleu
            value: 20.2
          - name: chr-F
            type: chrf
            value: 0.5238
      - task:
          name: Translation fra-lav
          type: translation
          args: fra-lav
        dataset:
          name: flores101-devtest
          type: flores_101
          args: fra lav devtest
        metrics:
          - name: BLEU
            type: bleu
            value: 23
          - name: chr-F
            type: chrf
            value: 0.5339
      - task:
          name: Translation fra-lit
          type: translation
          args: fra-lit
        dataset:
          name: flores101-devtest
          type: flores_101
          args: fra lit devtest
        metrics:
          - name: BLEU
            type: bleu
            value: 21.1
          - name: chr-F
            type: chrf
            value: 0.53595
      - task:
          name: Translation glg-lav
          type: translation
          args: glg-lav
        dataset:
          name: flores101-devtest
          type: flores_101
          args: glg lav devtest
        metrics:
          - name: BLEU
            type: bleu
            value: 20.7
          - name: chr-F
            type: chrf
            value: 0.51043
      - task:
          name: Translation glg-lit
          type: translation
          args: glg-lit
        dataset:
          name: flores101-devtest
          type: flores_101
          args: glg lit devtest
        metrics:
          - name: BLEU
            type: bleu
            value: 19.9
          - name: chr-F
            type: chrf
            value: 0.51854
      - task:
          name: Translation ita-lav
          type: translation
          args: ita-lav
        dataset:
          name: flores101-devtest
          type: flores_101
          args: ita lav devtest
        metrics:
          - name: BLEU
            type: bleu
            value: 19.6
          - name: chr-F
            type: chrf
            value: 0.51065
      - task:
          name: Translation ita-lit
          type: translation
          args: ita-lit
        dataset:
          name: flores101-devtest
          type: flores_101
          args: ita lit devtest
        metrics:
          - name: BLEU
            type: bleu
            value: 17.4
          - name: chr-F
            type: chrf
            value: 0.51309
      - task:
          name: Translation por-lav
          type: translation
          args: por-lav
        dataset:
          name: flores101-devtest
          type: flores_101
          args: por lav devtest
        metrics:
          - name: BLEU
            type: bleu
            value: 22.9
          - name: chr-F
            type: chrf
            value: 0.53493
      - task:
          name: Translation por-lit
          type: translation
          args: por-lit
        dataset:
          name: flores101-devtest
          type: flores_101
          args: por lit devtest
        metrics:
          - name: BLEU
            type: bleu
            value: 21.8
          - name: chr-F
            type: chrf
            value: 0.53821
      - task:
          name: Translation spa-lav
          type: translation
          args: spa-lav
        dataset:
          name: flores101-devtest
          type: flores_101
          args: spa lav devtest
        metrics:
          - name: BLEU
            type: bleu
            value: 17.4
          - name: chr-F
            type: chrf
            value: 0.4929
      - task:
          name: Translation spa-lit
          type: translation
          args: spa-lit
        dataset:
          name: flores101-devtest
          type: flores_101
          args: spa lit devtest
        metrics:
          - name: BLEU
            type: bleu
            value: 16.2
          - name: chr-F
            type: chrf
            value: 0.49836
      - task:
          name: Translation ita-lit
          type: translation
          args: ita-lit
        dataset:
          name: tatoeba-test-v2021-08-07
          type: tatoeba_mt
          args: ita-lit
        metrics:
          - name: BLEU
            type: bleu
            value: 40.9
          - name: chr-F
            type: chrf
            value: 0.6764
      - task:
          name: Translation spa-lit
          type: translation
          args: spa-lit
        dataset:
          name: tatoeba-test-v2021-08-07
          type: tatoeba_mt
          args: spa-lit
        metrics:
          - name: BLEU
            type: bleu
            value: 45.9
          - name: chr-F
            type: chrf
            value: 0.68805

opus-mt-tc-big-itc-bat

Table of Contents

Model Details

Neural machine translation model for translating from Italic languages (itc) to Baltic languages (bat).

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:

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. >>lav<<

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 = [
    ">>lit<< Els gats són complexos individus.",
    ">>sgs<< No."
]

model_name = "pytorch-models/opus-mt-tc-big-itc-bat"
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:
#     Katės yra sudėtingi individai.
#     no no no no no no no no no no no no no no no no no no no no no

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-bat")
print(pipe(">>lit<< Els gats són complexos individus."))

# expected output: Katės yra sudėtingi individai.

Training

Evaluation

langpair testset chr-F BLEU #sent #words
ita-lit tatoeba-test-v2021-08-07 0.67640 40.9 224 1321
spa-lit tatoeba-test-v2021-08-07 0.68805 45.9 454 2352
cat-lav flores101-devtest 0.52215 21.9 1012 22092
cat-lit flores101-devtest 0.52380 20.2 1012 20695
fra-lav flores101-devtest 0.53390 23.0 1012 22092
fra-lit flores101-devtest 0.53595 21.1 1012 20695
glg-lav flores101-devtest 0.51043 20.7 1012 22092
glg-lit flores101-devtest 0.51854 19.9 1012 20695
ita-lav flores101-devtest 0.51065 19.6 1012 22092
ita-lit flores101-devtest 0.51309 17.4 1012 20695
por-lav flores101-devtest 0.53493 22.9 1012 22092
por-lit flores101-devtest 0.53821 21.8 1012 20695
spa-lav flores101-devtest 0.49290 17.4 1012 22092
spa-lit flores101-devtest 0.49836 16.2 1012 20695

Citation Information

@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 00:04:44 EEST 2022
  • port machine: LM0-400-22516.local