Attend the Women in ML Symposium on December 7 Register now

mtnt

  • Description:

MTNT: Machine Translation of Noisy Text

FeaturesDict({
    'dst': Text(shape=(), dtype=object),
    'src': Text(shape=(), dtype=object),
})
  • Feature documentation:
Feature Class Shape Dtype Description
FeaturesDict
dst Text object
src Text object
@InProceedings{michel2018mtnt,
  author    = {Michel, Paul  and  Neubig, Graham},
  title     = {MTNT: A Testbed for Machine Translation of Noisy Text},
  booktitle = {Proceedings of the 2018 Conference on Empirical Methods in Natural Language Processing}
}

mtnt/en-fr (default config)

  • Dataset size: 11.33 MiB

  • Splits:

Split Examples
'test' 1,020
'train' 35,692
'valid' 811

mtnt/en-ja

  • Dataset size: 4.41 MiB

  • Splits:

Split Examples
'test' 946
'train' 5,746
'valid' 892

mtnt/fr-en

  • Dataset size: 8.28 MiB

  • Splits:

Split Examples
'test' 1,022
'train' 18,942
'valid' 876