@inproceedings{zouhar-2021-sampling,
    title = "Sampling and Filtering of Neural Machine Translation Distillation Data",
    author = "Zouhar, Vil{\'e}m",
    editor = "Durmus, Esin  and
      Gupta, Vivek  and
      Liu, Nelson  and
      Peng, Nanyun  and
      Su, Yu",
    booktitle = "Proceedings of the 2021 Conference of the North American Chapter of the Association for Computational Linguistics: Student Research Workshop",
    month = jun,
    year = "2021",
    address = "Online",
    publisher = "Association for Computational Linguistics",
    url = "https://aclanthology.org/2021.naacl-srw.1",
    doi = "10.18653/v1/2021.naacl-srw.1",
    pages = "1--8",
    abstract = "In most of neural machine translation distillation or stealing scenarios, the highest-scoring hypothesis of the target model (teacher) is used to train a new model (student). If reference translations are also available, then better hypotheses (with respect to the references) can be oversampled and poor hypotheses either removed or undersampled. This paper explores the sampling method landscape (pruning, hypothesis oversampling and undersampling, deduplication and their combination) with English to Czech and English to German MT models using standard MT evaluation metrics. We show that careful oversampling and combination with the original data leads to better performance when compared to training only on the original or synthesized data or their direct combination.",
}