Task-specific Compression for Multi-task Language Models using Attribution-based Pruning

被引:0
|
作者
Yang, Nakyeong [1 ]
Jang, Yunah [1 ]
Lee, Hwanhee [2 ]
Jung, Seohyeong [3 ]
Jung, Kyomin [1 ]
机构
[1] Seoul Natl Univ, Seoul, South Korea
[2] Chung Ang Univ, Seoul, South Korea
[3] Hyundai Motor Grp & 42dot Inc, Seoul, South Korea
关键词
D O I
暂无
中图分类号
TP18 [人工智能理论];
学科分类号
081104 ; 0812 ; 0835 ; 1405 ;
摘要
Multi-task language models show outstanding performance for various natural language understanding tasks with only a single model. However, these language models utilize an unnecessarily large number of model parameters, even when used only for a specific task. This paper proposes a novel training-free compression method for multi-task language models using a pruning method. Specifically, we use an attribution method to determine which neurons are essential for performing a specific task. We task-specifically prune unimportant neurons and leave only task-specific parameters. Furthermore, we extend our method to be applicable in low-resource and unsupervised settings. Since our compression method is training-free, it uses few computing resources and does not destroy the pre-trained knowledge of language models. Experimental results on the six widely-used datasets show that our proposed pruning method significantly outperforms baseline pruning methods. In addition, we demonstrate that our method preserves performance even in an unseen domain setting.
引用
收藏
页码:594 / 604
页数:11
相关论文
共 50 条
  • [1] Task-Specific Pruning: Efficient Parameter Reduction in Multi-task Object Detection Models
    Ke, Wei-Hsun
    Tseng, Yu-Wen
    Cheng, Wen-Huang
    2023 ASIA PACIFIC SIGNAL AND INFORMATION PROCESSING ASSOCIATION ANNUAL SUMMIT AND CONFERENCE, APSIPA ASC, 2023, : 1712 - 1717
  • [2] Finding Task-specific Subnetworks in Multi-task Spoken Language Understanding Model
    Futami, Hayato
    Arora, Siddhant
    Kashiwagi, Yosuke
    Tsunoo, Emiru
    Watanabe, Shinji
    INTERSPEECH 2024, 2024, : 802 - 806
  • [3] A Kernel Approach to Multi-Task Learning with Task-Specific Kernels
    Wu, Wei
    Li, Hang
    Hu, Yun-Hua
    Jin, Rong
    JOURNAL OF COMPUTER SCIENCE AND TECHNOLOGY, 2012, 27 (06) : 1289 - 1301
  • [4] A Kernel Approach to Multi-Task Learning with Task-Specific Kernels
    武威
    李航
    胡云华
    金榕
    Journal of Computer Science & Technology, 2012, 27 (06) : 1289 - 1301
  • [5] A Kernel Approach to Multi-Task Learning with Task-Specific Kernels
    Wei Wu
    Hang Li
    Yun-Hua Hu
    Rong Jin
    Journal of Computer Science and Technology, 2012, 27 : 1289 - 1301
  • [6] AutoDistiller: An Automatic Compression Method for Multi-task Language Models
    Wang, Hongsheng
    Xiao, Geyang
    Liang, Yuan
    2022 41ST CHINESE CONTROL CONFERENCE (CCC), 2022, : 2410 - 2415
  • [7] Deep Task-specific Bottom Representation Network for Multi-Task Recommendation
    Liu, Qi
    Zhou, Zhilong
    Jiang, Gangwei
    Ge, Tiezheng
    Lian, Defu
    PROCEEDINGS OF THE 32ND ACM INTERNATIONAL CONFERENCE ON INFORMATION AND KNOWLEDGE MANAGEMENT, CIKM 2023, 2023, : 1637 - 1646
  • [8] Task-specific abilities in multi-task principal-agent relationships
    Thiele, Veikko
    LABOUR ECONOMICS, 2010, 17 (04) : 690 - 698
  • [9] Channel Attention-Based Method for Searching Task-Specific Multi-Task Network Structures
    Ye, Songtao
    Zheng, Saisai
    Xiao, Yizhang
    PROCEEDINGS OF THE 2024 27 TH INTERNATIONAL CONFERENCE ON COMPUTER SUPPORTED COOPERATIVE WORK IN DESIGN, CSCWD 2024, 2024, : 562 - 569
  • [10] IMAGE COMPRESSION BASED ON TASK-SPECIFIC INFORMATION
    Pu, Lingling
    Marcellin, Michael W.
    Bilgin, Ali
    Ashok, Amit
    2014 IEEE INTERNATIONAL CONFERENCE ON IMAGE PROCESSING (ICIP), 2014, : 4817 - 4821