Abstract
To reduce the inference cost of large language models, model compression is increasingly used to create smaller scalable models. However, little is known about their robustness to minority subgroups defined by the labels and attributes of a dataset. In this paper, we investigate the effects of 18 different compression methods and settings on the subgroup robustness of BERT language models. We show that worst-group performance does not depend on model size alone, but also on the compression method used. Additionally, we find that model compression does not always worsen the performance on minority subgroups. Altogether, our analysis serves to further research into the subgroup robustness of model compression.- Anthology ID:
- 2023.emnlp-main.983
- Volume:
- Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing
- Month:
- December
- Year:
- 2023
- Address:
- Singapore
- Editors:
- Houda Bouamor, Juan Pino, Kalika Bali
- Venue:
- EMNLP
- SIG:
- Publisher:
- Association for Computational Linguistics
- Note:
- Pages:
- 15859–15868
- Language:
- URL:
- https://aclanthology.org/2023.emnlp-main.983
- DOI:
- 10.18653/v1/2023.emnlp-main.983
- Cite (ACL):
- Leonidas Gee, Andrea Zugarini, and Novi Quadrianto. 2023. Are Compressed Language Models Less Subgroup Robust?. In Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing, pages 15859–15868, Singapore. Association for Computational Linguistics.
- Cite (Informal):
- Are Compressed Language Models Less Subgroup Robust? (Gee et al., EMNLP 2023)
- PDF:
- https://preview.aclanthology.org/corrections-2024-05/2023.emnlp-main.983.pdf