Memorisation versus Generalisation in Pre-trained Language Models

被引:0
|
作者
Tanzer, Michael [1 ]
Ruder, Sebastian [2 ]
Rei, Marek [1 ]
机构
[1] Imperial Coll London, London, England
[2] Google Res, Mountain View, CA USA
来源
PROCEEDINGS OF THE 60TH ANNUAL MEETING OF THE ASSOCIATION FOR COMPUTATIONAL LINGUISTICS (ACL 2022), VOL 1: (LONG PAPERS) | 2022年
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
State-of-the-art pre-trained language models have been shown to memorise facts and perform well with limited amounts of training data. To gain a better understanding of how these models learn, we study their generalisation and memorisation capabilities in noisy and low-resource scenarios. We find that the training of these models is almost unaffected by label noise and that it is possible to reach near-optimal results even on extremely noisy datasets. However, our experiments also show that they mainly learn from high-frequency patterns and largely fail when tested on low-resource tasks such as few-shot learning and rare entity recognition. To mitigate such limitations, we propose an extension based on prototypical networks that improves performance in low-resource named entity recognition tasks.
引用
收藏
页码:7564 / 7578
页数:15
相关论文
共 50 条
  • [1] Pre-Trained Language Models and Their Applications
    Wang, Haifeng
    Li, Jiwei
    Wu, Hua
    Hovy, Eduard
    Sun, Yu
    ENGINEERING, 2023, 25 : 51 - 65
  • [2] Annotating Columns with Pre-trained Language Models
    Suhara, Yoshihiko
    Li, Jinfeng
    Li, Yuliang
    Zhang, Dan
    Demiralp, Cagatay
    Chen, Chen
    Tan, Wang-Chiew
    PROCEEDINGS OF THE 2022 INTERNATIONAL CONFERENCE ON MANAGEMENT OF DATA (SIGMOD '22), 2022, : 1493 - 1503
  • [3] LaoPLM: Pre-trained Language Models for Lao
    Lin, Nankai
    Fu, Yingwen
    Yang, Ziyu
    Chen, Chuwei
    Jiang, Shengyi
    LREC 2022: THIRTEEN INTERNATIONAL CONFERENCE ON LANGUAGE RESOURCES AND EVALUATION, 2022, : 6506 - 6512
  • [4] Knowledge Rumination for Pre-trained Language Models
    Yao, Yunzhi
    Wang, Peng
    Mao, Shengyu
    Tan, Chuanqi
    Huang, Fei
    Chen, Huajun
    Zhang, Ningyu
    2023 CONFERENCE ON EMPIRICAL METHODS IN NATURAL LANGUAGE PROCESSING, EMNLP 2023, 2023, : 3387 - 3404
  • [5] Deciphering Stereotypes in Pre-Trained Language Models
    Ma, Weicheng
    Scheible, Henry
    Wang, Brian
    Veeramachaneni, Goutham
    Chowdhary, Pratim
    Sung, Alan
    Koulogeorge, Andrew
    Wang, Lili
    Yang, Diyi
    Vosoughi, Soroush
    2023 CONFERENCE ON EMPIRICAL METHODS IN NATURAL LANGUAGE PROCESSING (EMNLP 2023), 2023, : 11328 - 11345
  • [6] PhoBERT: Pre-trained language models for Vietnamese
    Dat Quoc Nguyen
    Anh Tuan Nguyen
    FINDINGS OF THE ASSOCIATION FOR COMPUTATIONAL LINGUISTICS, EMNLP 2020, 2020, : 1037 - 1042
  • [7] HinPLMs: Pre-trained Language Models for Hindi
    Huang, Xixuan
    Lin, Nankai
    Li, Kexin
    Wang, Lianxi
    Gan, Suifu
    2021 INTERNATIONAL CONFERENCE ON ASIAN LANGUAGE PROCESSING (IALP), 2021, : 241 - 246
  • [8] Evaluating Commonsense in Pre-Trained Language Models
    Zhou, Xuhui
    Zhang, Yue
    Cui, Leyang
    Huang, Dandan
    THIRTY-FOURTH AAAI CONFERENCE ON ARTIFICIAL INTELLIGENCE, THE THIRTY-SECOND INNOVATIVE APPLICATIONS OF ARTIFICIAL INTELLIGENCE CONFERENCE AND THE TENTH AAAI SYMPOSIUM ON EDUCATIONAL ADVANCES IN ARTIFICIAL INTELLIGENCE, 2020, 34 : 9733 - 9740
  • [9] Knowledge Inheritance for Pre-trained Language Models
    Qin, Yujia
    Lin, Yankai
    Yi, Jing
    Zhang, Jiajie
    Han, Xu
    Zhang, Zhengyan
    Su, Yusheng
    Liu, Zhiyuan
    Li, Peng
    Sun, Maosong
    Zhou, Jie
    NAACL 2022: THE 2022 CONFERENCE OF THE NORTH AMERICAN CHAPTER OF THE ASSOCIATION FOR COMPUTATIONAL LINGUISTICS: HUMAN LANGUAGE TECHNOLOGIES, 2022, : 3921 - 3937
  • [10] Code Execution with Pre-trained Language Models
    Liu, Chenxiao
    Lu, Shuai
    Chen, Weizhu
    Jiang, Daxin
    Svyatkovskiy, Alexey
    Fu, Shengyu
    Sundaresan, Neel
    Duan, Nan
    FINDINGS OF THE ASSOCIATION FOR COMPUTATIONAL LINGUISTICS, ACL 2023, 2023, : 4984 - 4999