@inproceedings{velayuthan-etal-2025-encoder,
    title = "Encoder-Aware Sequence-Level Knowledge Distillation for Low-Resource Neural Machine Translation",
    author = "Velayuthan, Menan  and
      De Silva, Nisansa  and
      Ranathunga, Surangika",
    editor = "Ojha, Atul Kr.  and
      Liu, Chao-hong  and
      Vylomova, Ekaterina  and
      Pirinen, Flammie  and
      Washington, Jonathan  and
      Oco, Nathaniel  and
      Zhao, Xiaobing",
    booktitle = "Proceedings of the Eighth Workshop on Technologies for Machine Translation of Low-Resource Languages (LoResMT 2025)",
    month = may,
    year = "2025",
    address = "Albuquerque, New Mexico, U.S.A.",
    publisher = "Association for Computational Linguistics",
    url = "https://preview.aclanthology.org/fix-sig-urls/2025.loresmt-1.15/",
    pages = "161--170",
    ISBN = "979-8-89176-230-5",
    abstract = "Domain adaptation in Neural Machine Translation (NMT) is commonly achieved through fine-tuning, but this approach becomes inefficient as the number of domains increases. Knowledge distillation (KD) provides a scalable alternative by training a compact model on distilled data from a larger model. However, we hypothesize that vanilla sequence-level KD primarily distills the decoder while neglecting encoder knowledge, leading to suboptimal knowledge transfer and limiting its effectiveness in low-resource settings, where both data and computational resources are constrained. To address this, we propose an improved sequence-level KD method that enhances encoder knowledge transfer through a cosine-based alignment loss. Our approach first trains a large model on a mixed-domain dataset and generates a Distilled Mixed Dataset (DMD). A small model is then trained on this dataset via sequence-level KD with encoder alignment. Experiments in a low-resource setting validate our hypothesis, demonstrating that our approach outperforms vanilla sequence-level KD, improves generalization to out-of-domain data, and facilitates efficient domain adaptation while reducing model size and computational cost."
}