GLOSS-MT: Controllable Low-Resource Machine Translation with an Editable Dictionary

Key points

• GLOSS-MT looks up the source's noun phrases in an offline English-Twi dictionary, replaces every phrase found by its Twi translation in place, and fine-tunes a translation model once to translate these mixed inputs.1
• We apply the same data and pipeline to NLLB-200 (600M, 1.3B, 3.3B), OPUS-MT en-tw and TranslateGemma 4B.1
• A direct controllability test on human-translated Bible passages shows that the three fine-tunes tested write 79-97% of the dictionary terms, and that replacing every dictionary entry by a different term changes the output accordingly (70-97% of the new terms appear) with no retraining.1
• We release the models, data and code.1

Sources (1)

• [1]GLOSS-MT: Controllable Low-Resource Machine Translation with an Editable Dictionary
  arXiv (AI, ML, NLP, CV, robotics, multi-agent) · Oct 17, 02:50 PM

  Adapting a pretrained machine translation (MT) model to new terminology usually requires new parallel data and another round of fine-tuning, which is costly for low-resource languages.

  GLOSS-MT looks up the source's noun phrases in an offline English-Twi dictionary, replaces every phrase found by its Twi translation in place, and fine-tunes a translation model once to translate these mixed inputs.

Extractive summary: sentences quoted from the sources.