Measuring and Improving Consistency in Pretrained Language Models
Abstrak
Abstract Consistency of a model—that is, the invariance of its behavior under meaning-preserving alternations in its input—is a highly desirable property in natural language processing. In this paper we study the question: Are Pretrained Language Models (PLMs) consistent with respect to factual knowledge? To this end, we create ParaRel🤘, a high-quality resource of cloze-style query English paraphrases. It contains a total of 328 paraphrases for 38 relations. Using ParaRel🤘, we show that the consistency of all PLMs we experiment with is poor— though with high variance between relations. Our analysis of the representational spaces of PLMs suggests that they have a poor structure and are currently not suitable for representing knowledge robustly. Finally, we propose a method for improving model consistency and experimentally demonstrate its effectiveness.1
Topik & Kata Kunci
Penulis (7)
Yanai Elazar
Nora Kassner
Shauli Ravfogel
Abhilasha Ravichander
E. Hovy
Hinrich Schütze
Yoav Goldberg
Akses Cepat
- Tahun Terbit
- 2021
- Bahasa
- en
- Total Sitasi
- 460×
- Sumber Database
- Semantic Scholar
- DOI
- 10.1162/tacl_a_00410
- Akses
- Open Access ✓