@inproceedings{wang-mcallester-2020-fly,
    title = "On-The-Fly Information Retrieval Augmentation for Language Models",
    author = "Wang, Hai  and
      McAllester, David",
    editor = "Bonial, Claire  and
      Caselli, Tommaso  and
      Chaturvedi, Snigdha  and
      Clark, Elizabeth  and
      Huang, Ruihong  and
      Iyyer, Mohit  and
      Jaimes, Alejandro  and
      Ji, Heng  and
      Martin, Lara J.  and
      Miller, Ben  and
      Mitamura, Teruko  and
      Peng, Nanyun  and
      Tetreault, Joel",
    booktitle = "Proceedings of the First Joint Workshop on Narrative Understanding, Storylines, and Events",
    month = jul,
    year = "2020",
    address = "Online",
    publisher = "Association for Computational Linguistics",
    url = "https://preview.aclanthology.org/landing_page/2020.nuse-1.14/",
    doi = "10.18653/v1/2020.nuse-1.14",
    pages = "114--119",
    abstract = "Here we experiment with the use of information retrieval as an augmentation for pre-trained language models. The text corpus used in information retrieval can be viewed as form of episodic memory which grows over time. By augmenting GPT 2.0 with information retrieval we achieve a zero shot 15{\%} relative reduction in perplexity on Gigaword corpus without any re-training. We also validate our IR augmentation on an event co-reference task."
}