Probing BERT's priors with serial reproduction chains

Takateru Yamakoshi, Thomas L. Griffiths, Robert D. Hawkins

Research output: Chapter in Book/Report/Conference proceedingConference contribution

3 Scopus citations

Abstract

Sampling is a promising bottom-up method for exposing what generative models have learned about language, but it remains unclear how to generate representative samples from popular masked language models (MLMs) like BERT. The MLM objective yields a dependency network with no guarantee of consistent conditional distributions, posing a problem for naive approaches. Drawing from theories of iterated learning in cognitive science, we explore the use of serial reproduction chains to sample from BERT's priors. In particular, we observe that a unique and consistent estimator of the ground-truth joint distribution is given by a Generative Stochastic Network (GSN) sampler, which randomly selects which token to mask and reconstruct on each step. We show that the lexical and syntactic statistics of sentences from GSN chains closely match the ground-truth corpus distribution and perform better than other methods in a large corpus of naturalness judgments. Our findings establish a firmer theoretical foundation for bottom-up probing and highlight richer deviations from human priors.

Original languageEnglish (US)
Title of host publicationACL 2022 - 60th Annual Meeting of the Association for Computational Linguistics, Findings of ACL 2022
EditorsSmaranda Muresan, Preslav Nakov, Aline Villavicencio
PublisherAssociation for Computational Linguistics (ACL)
Pages3977-3992
Number of pages16
ISBN (Electronic)9781955917254
StatePublished - 2022
Event60th Annual Meeting of the Association for Computational Linguistics, ACL 2022 - Dublin, Ireland
Duration: May 22 2022May 27 2022

Publication series

NameProceedings of the Annual Meeting of the Association for Computational Linguistics
ISSN (Print)0736-587X

Conference

Conference60th Annual Meeting of the Association for Computational Linguistics, ACL 2022
Country/TerritoryIreland
CityDublin
Period5/22/225/27/22

All Science Journal Classification (ASJC) codes

  • Computer Science Applications
  • Linguistics and Language
  • Language and Linguistics

Fingerprint

Dive into the research topics of 'Probing BERT's priors with serial reproduction chains'. Together they form a unique fingerprint.

Cite this