EMNLP 2022industry4 citations

PILE: Pairwise Iterative Logits Ensemble for Multi-Teacher Labeled Distillation

Lianshang Cai, Linhao Zhang, Dehong Ma, Jun Fan, Daiting Shi, Yi Wu, Zhicong Cheng, Simiu Gu

Abstract

Pre-trained language models have become a crucial part of ranking systems and achieved very impressive effects recently. To maintain high performance while keeping efficient computations, knowledge distillation is widely used. In this paper, we focus on two key questions in knowledge distillation for ranking models: 1) how to ensemble knowledge from multi-teacher; 2) how to utilize the label information of data in the distillation process. We propose a unified algorithm called Pairwise Iterative Logits Ensemble (PILE) to tackle these two questions simultaneously. PILE ensembles multi-teacher logits supervised by label information in an iterative way and achieved competitive performance in both offline and online experiments. The proposed method has been deployed in a real-world commercial search system.

BibTeX
@inproceedings{cai-etal-2022-pile,
    title = "{PILE}: Pairwise Iterative Logits Ensemble for Multi-Teacher Labeled Distillation",
    author = "Cai, Lianshang  and
      Zhang, Linhao  and
      Ma, Dehong  and
      Fan, Jun  and
      Shi, Daiting  and
      Wu, Yi  and
      Cheng, Zhicong  and
      Gu, Simiu  and
      Yin, Dawei",
    editor = "Li, Yunyao  and
      Lazaridou, Angeliki",
    booktitle = "Proceedings of the 2022 Conference on Empirical Methods in Natural Language Processing: Industry Track",
    month = dec,
    year = "2022",
    address = "Abu Dhabi, UAE",
    publisher = "Association for Computational Linguistics",
    url = "https://aclanthology.org/2022.emnlp-industry.60/",
    doi = "10.18653/v1/2022.emnlp-industry.60",
    pages = "587--595"
}
PILE: Pairwise Iterative Logits Ensemble for Multi-Teacher Labeled Distillation · EMNLP 2022