Recent studies have shown that transformer models like BERT rely on number information encoded in their representations of sentences’ subjects and head verbs when performing subject-verb agreement. However, probing experiments suggest that subject number is also encoded in the representations of all words in such sentences. In this paper, we use causal interventions to show that BERT only uses the subject plurality information encoded in its representations of the subject and words that agree with it in number. We also demonstrate that current probing metrics are unable to determine which words’ representations contain functionally relevant information. This both provides a revised view of subject-verb agreement in language models, and suggests potential pitfalls for current probe usage and evaluation.

The Functional Relevance of Probed Information: A Case Study / Hanna, Michael; Zamparelli, Roberto; Mareček, David. - ELETTRONICO. - (2023), pp. 835-848. (Intervento presentato al convegno 17th Conference of the European Chapter of the Association for Computational Linguistics, EACL 2023 tenutosi a Dubrovnik, Croatia nel 2 maggio-6 maggio 2023) [10.18653/V1/2023.EACL-MAIN.58].

The Functional Relevance of Probed Information: A Case Study

Zamparelli, Roberto;
2023-01-01

Abstract

Recent studies have shown that transformer models like BERT rely on number information encoded in their representations of sentences’ subjects and head verbs when performing subject-verb agreement. However, probing experiments suggest that subject number is also encoded in the representations of all words in such sentences. In this paper, we use causal interventions to show that BERT only uses the subject plurality information encoded in its representations of the subject and words that agree with it in number. We also demonstrate that current probing metrics are unable to determine which words’ representations contain functionally relevant information. This both provides a revised view of subject-verb agreement in language models, and suggests potential pitfalls for current probe usage and evaluation.
2023
Proceedings of the 17th Conference of the European Chapter of the Association for Computational Linguistics
Dubrovnik, Croatia
Association for Computational Linguistics (ACL)
978-1-959429-44-9
Hanna, Michael; Zamparelli, Roberto; Mareček, David
The Functional Relevance of Probed Information: A Case Study / Hanna, Michael; Zamparelli, Roberto; Mareček, David. - ELETTRONICO. - (2023), pp. 835-848. (Intervento presentato al convegno 17th Conference of the European Chapter of the Association for Computational Linguistics, EACL 2023 tenutosi a Dubrovnik, Croatia nel 2 maggio-6 maggio 2023) [10.18653/V1/2023.EACL-MAIN.58].
File in questo prodotto:
File Dimensione Formato  
2023.eacl-main.58.pdf

accesso aperto

Descrizione: The Functional Relevance of Probed Information: A Case Study
Tipologia: Versione editoriale (Publisher’s layout)
Licenza: Creative commons
Dimensione 1.02 MB
Formato Adobe PDF
1.02 MB Adobe PDF Visualizza/Apri

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/11572/400476
Citazioni
  • ???jsp.display-item.citation.pmc??? ND
  • Scopus 2
  • ???jsp.display-item.citation.isi??? ND
social impact