Language Models Are Poor Learners of Directional Inference
Li, Tianyi; Hosseini, Mohammad Javad; Weber, Sabine; u. a. (2025): Language Models Are Poor Learners of Directional Inference, in: Bamberg: Otto-Friedrich-Universität, S. 903–921.
Details
- Fakultät/Lehrstuhl
- Autorin(nen)/Autor(en)
- Li, TianyiHosseini, Mohammad JavadSteedman, Mark
- Verlagsangabe
- Otto-Friedrich-Universität : Bamberg
- Erscheinungsjahr
- 2025
- Quelle/Weitere Ausgabe
- Yoav Goldberg, Zornitsa Kozareva, und Yue Zhang (Hrsg.), Findings of the Association for Computational Linguistics: EMNLP 2022, Association for Computational Linguistics, 2022, S. 903–921
- Jahr der Erstpublikation
- 2022
- Sprache(n)
- Englisch
- Lizenz
- URN
- Abstract
We examine LMs’ competence of directional predicate entailments by supervised fine-tuning with prompts. Our analysis shows that contrary to their apparent success on standard NLI, LMs show limited ability to learn such directional inference; moreover, existing datasets fail to test directionality, and/or are infested by artefacts that can be learnt as proxy for entailments, yielding over-optimistic results. In response, we present BoOQA (Boolean Open QA), a robust multi-lingual evaluation benchmark for directional predicate entailments, extrinsic to existing training sets. On BoOQA, we establish baselines and show evidence of existing LM-prompting models being incompetent directional entailment learners, in contrast to entailment graphs, however limited by sparsity.
- Freie Schlagwort(e)
- Language Models
- Peer Reviewed
- ja
- Internationale Verbreitung
- ja
- Open-Access-Zeitschrift
- ja
- Dokumenttyp
- Konferenzveröffentlichung
- URI
- Datum der Freischaltung
- 10.11.2025