Abstract
In this work, we propose a method that combines two popular research areas by injecting linguistic structures into pre-trained language models in the parameter-efficient fine-tuning (PEFT) setting. In our approach, parallel adapter modules encoding different linguistic structures are combined using a novel Mixture-of-Linguistic-Experts architecture, where Gumbel-Softmax gates are used to determine the importance of these modules at each layer of the model. To reduce the number of parameters, we first train the model for a fixed small number of steps before pruning the experts based on their important scores. Our experiment results with three different pre-trained models show that our approach can outperform state-of-the-art PEFT methods with a comparable number of parameters. In addition, we provide additional analysis to examine the experts selected by each model at each layer to provide insights for future studies.- Anthology ID:
- 2023.findings-emnlp.634
- Volume:
- Findings of the Association for Computational Linguistics: EMNLP 2023
- Month:
- December
- Year:
- 2023
- Address:
- Singapore
- Editors:
- Houda Bouamor, Juan Pino, Kalika Bali
- Venue:
- Findings
- SIG:
- Publisher:
- Association for Computational Linguistics
- Note:
- Pages:
- 9456–9469
- Language:
- URL:
- https://aclanthology.org/2023.findings-emnlp.634
- DOI:
- 10.18653/v1/2023.findings-emnlp.634
- Cite (ACL):
- Raymond Li, Gabriel Murray, and Giuseppe Carenini. 2023. Mixture-of-Linguistic-Experts Adapters for Improving and Interpreting Pre-trained Language Models. In Findings of the Association for Computational Linguistics: EMNLP 2023, pages 9456–9469, Singapore. Association for Computational Linguistics.
- Cite (Informal):
- Mixture-of-Linguistic-Experts Adapters for Improving and Interpreting Pre-trained Language Models (Li et al., Findings 2023)
- PDF:
- https://preview.aclanthology.org/nschneid-patch-5/2023.findings-emnlp.634.pdf