Beyond the Bias: Unveiling the Quality of Implicit Causality Prompt Continuations in Language Models

Judith Sieker, Oliver Bott, Torgrim Solstad, Sina Zarrieß

Abstract
Recent studies have used human continuations of Implicit Causality (IC) prompts collected in linguistic experiments to evaluate discourse understanding in large language models (LLMs), focusing on the well-known IC coreference bias in the LLMs’ predictions of the next word following the prompt. In this study, we investigate how continuations of IC prompts can be used to evaluate the text generation capabilities of LLMs in a linguistically controlled setting. We conduct an experiment using two open-source GPT-based models, employing human evaluation to assess different aspects of continuation quality. Our findings show that LLMs struggle in particular with generating coherent continuations in this rather simple setting, indicating a lack of discourse knowledge beyond the well-known IC bias. Our results also suggest that a bias congruent continuation does not necessarily equate to a higher continuation quality. Furthermore, our study draws upon insights from the Uniform Information Density hypothesis, testing different prompt modifications and decoding procedures and showing that sampling-based methods are particularly sensitive to the information density of the prompts.
Anthology ID:
2023.inlg-main.15
Volume:
Proceedings of the 16th International Natural Language Generation Conference
Month:
September
Year:
2023
Address:
Prague, Czechia
Editors:
C. Maria Keet, Hung-Yi Lee, Sina Zarrieß
Venues:
INLG | SIGDIAL
SIG:
SIGGEN
Publisher:
Association for Computational Linguistics
Note:
Pages:
206–220
Language:
URL:
https://aclanthology.org/2023.inlg-main.15
DOI:
10.18653/v1/2023.inlg-main.15
Bibkey:
Cite (ACL):
Judith Sieker, Oliver Bott, Torgrim Solstad, and Sina Zarrieß. 2023. Beyond the Bias: Unveiling the Quality of Implicit Causality Prompt Continuations in Language Models. In Proceedings of the 16th International Natural Language Generation Conference, pages 206–220, Prague, Czechia. Association for Computational Linguistics.
Cite (Informal):
Beyond the Bias: Unveiling the Quality of Implicit Causality Prompt Continuations in Language Models (Sieker et al., INLG-SIGDIAL 2023)
Copy Citation:
PDF:
https://preview.aclanthology.org/nschneid-patch-5/2023.inlg-main.15.pdf
PDF Search