@inproceedings{lin-etal-2025-self,
    title = "Self-Vocabularizing Training for Neural Machine Translation",
    author = "Lin, Pin-Jie  and
      Chang, Ernie  and
      Shi, Yangyang  and
      Chandra, Vikas",
    editor = "Ebrahimi, Abteen  and
      Haider, Samar  and
      Liu, Emmy  and
      Haider, Sammar  and
      Leonor Pacheco, Maria  and
      Wein, Shira",
    booktitle = "Proceedings of the 2025 Conference of the Nations of the Americas Chapter of the Association for Computational Linguistics: Human Language Technologies (Volume 4: Student Research Workshop)",
    month = apr,
    year = "2025",
    address = "Albuquerque, USA",
    publisher = "Association for Computational Linguistics",
    url = "https://preview.aclanthology.org/moar-dois/2025.naacl-srw.16/",
    doi = "10.18653/v1/2025.naacl-srw.16",
    pages = "171--177",
    ISBN = "979-8-89176-192-6",
    abstract = "Past vocabulary learning techniques identify relevant vocabulary before training, relying on statistical and entropy-based assumptions that largely neglect the role of model training.Empirically, we observe that trained translation models are induced to use a byte-pair encoding (BPE) vocabulary subset distinct from the original BPE vocabulary, leading to performance improvements when retrained with the induced vocabulary.In this paper, we analyze this discrepancy in neural machine translation by examining vocabulary and entropy shifts during self-training{---}where each iteration generates a labeled dataset by pairing source sentences with the model{'}s predictions to define a new vocabulary.Building on these insights, we propose *self-vocabularizing training*, an iterative method that self-selects a smaller, more optimal vocabulary, yielding up to a 1.49 BLEU improvement.Moreover, we find that deeper model architectures lead to both an increase in unique token usage and a 6{--}8{\%} reduction in vocabulary size."
}