A Simple but Effective Pluggable Entity Lookup Table for Pre-trained Language Models
Abstract
Pre-trained language models (PLMs) cannot well recall rich factual knowledge of entities exhibited in large-scale corpora, especially those rare entities. In this paper, we propose to build a simple but effective Pluggable Entity Lookup Table (PELT) on demand by aggregating the entity's output representations of multiple occurrences in the corpora. PELT can be compatibly plugged as inputs to infuse supplemental entity knowledge into PLMs. Compared to previous knowledge-enhanced PLMs, PELT only requires 0.2%-5% pre-computation with capability of acquiring knowledge from out-of-domain corpora for domain adaptation scenario. The experiments on knowledge-related tasks demonstrate that our method, PELT, can flexibly and effectively transfer entity knowledge from related corpora into PLMs with different architectures.
Cite
@article{arxiv.2202.13392,
title = {A Simple but Effective Pluggable Entity Lookup Table for Pre-trained Language Models},
author = {Deming Ye and Yankai Lin and Peng Li and Maosong Sun and Zhiyuan Liu},
journal= {arXiv preprint arXiv:2202.13392},
year = {2022}
}
Comments
Accepted to ACL 2022. The code and models are available at https://github.com/thunlp/PELT