@inproceedings{zhou-etal-2024-govern,
    title = "{GOVERN}: Gradient Orientation Vote Ensemble for Multi-Teacher Reinforced Distillation",
    author = "Zhou, Wenjie  and
      Ding, Zhenxin  and
      Zhang, Xiaodong  and
      Shi, Haibo  and
      Wang, Junfeng  and
      Yin, Dawei",
    editor = "Dernoncourt, Franck  and
      Preo{\c{t}}iuc-Pietro, Daniel  and
      Shimorina, Anastasia",
    booktitle = "Proceedings of the 2024 Conference on Empirical Methods in Natural Language Processing: Industry Track",
    month = nov,
    year = "2024",
    address = "Miami, Florida, US",
    publisher = "Association for Computational Linguistics",
    url = "https://preview.aclanthology.org/add-emnlp-2024-awards/2024.emnlp-industry.120/",
    doi = "10.18653/v1/2024.emnlp-industry.120",
    pages = "1650--1658",
    abstract = "Pre-trained language models have become an integral component of question-answering systems, achieving remarkable performance. However, for practical deployment, it is crucial to perform knowledge distillation to maintain high performance while operating under computational constraints. In this paper, we address a key question: given the importance of unsupervised distillation for student model performance, how can knowledge from multiple teacher models be effectively ensemble during this stage without the guidance of labels? We propose a novel algorithm, GOVERN, to tackle this issue. GOVERN has demonstrated significant improvements in both offline and online experiments, enabling the student model to achieve results comparable to that of teacher ensembles. Our experiments show that GOVERN remarkably requires a mere 1{\%} of the ensemble method`s inference budget to achieve 99.5{\%} of performance. The proposed algorithm has been successfully deployed in a real-world commercial question-answering system, demonstrating its real-world applicability."
}