Abstract
It is unclear whether, how and where large pre-trained language models capture subtle linguistic traits like ambiguity, grammaticality and sentence complexity. We present results of automatic classification of these traits and compare their viability and patterns across representation types. We demonstrate that template-based datasets with surface-level artifacts should not be used for probing, careful comparisons with baselines should be done and that t-SNE plots should not be used to determine the presence of a feature among dense vectors representations. We also show how features might be highly localized in the layers for these models and get lost in the upper layers.- Anthology ID:
- 2022.blackboxnlp-1.4
- Volume:
- Proceedings of the Fifth BlackboxNLP Workshop on Analyzing and Interpreting Neural Networks for NLP
- Month:
- December
- Year:
- 2022
- Address:
- Abu Dhabi, United Arab Emirates (Hybrid)
- Editors:
- Jasmijn Bastings, Yonatan Belinkov, Yanai Elazar, Dieuwke Hupkes, Naomi Saphra, Sarah Wiegreffe
- Venue:
- BlackboxNLP
- SIG:
- Publisher:
- Association for Computational Linguistics
- Note:
- Pages:
- 40–50
- Language:
- URL:
- https://aclanthology.org/2022.blackboxnlp-1.4
- DOI:
- 10.18653/v1/2022.blackboxnlp-1.4
- Cite (ACL):
- Sunit Bhattacharya, Vilém Zouhar, and Ondrej Bojar. 2022. Sentence Ambiguity, Grammaticality and Complexity Probes. In Proceedings of the Fifth BlackboxNLP Workshop on Analyzing and Interpreting Neural Networks for NLP, pages 40–50, Abu Dhabi, United Arab Emirates (Hybrid). Association for Computational Linguistics.
- Cite (Informal):
- Sentence Ambiguity, Grammaticality and Complexity Probes (Bhattacharya et al., BlackboxNLP 2022)
- PDF:
- https://preview.aclanthology.org/nschneid-patch-4/2022.blackboxnlp-1.4.pdf