Dynamic Low-rank Estimation for Transformer-based Language Models
Ting Hua, Xiao Li, Shangqian Gao, Yen-Chang Hsu, Yilin Shen, Hongxia Jin
Abstract
Matrix decomposition methods, such as Singular Value Decomposition (SVD) and its importance-weighted variants, have been widely used for compressing Transformer-based language models. While importance-weighted decomposition methods alleviate the strong assumption of equal importance for each parameter in SVD, they still rely on two fundamental assumptions: 1) unchanged importance distribution during further fine-tuning, 2) equal importance across weight matrices in different layers. Furthermore, these methods necessitate a well-trained task-specific model as the starting point and require additional fine-tuning after compression. In this work, we proposed RankDyna, a matrix decomposition method that enables dynamic rank resource allocation among matrices across different layers during the training process. Starting from a general pre-trained model, RankDyna accomplishes the dual goals of compression and adaptation to the downstream task, all within a single round of fine-tuning. The extensive evaluations demonstrate that RankDyna can outperform current SOTA methods under various parameter budget levels, and the advantage of RankDyna is further enhanced with higher compression rates.- Anthology ID:
- 2023.findings-emnlp.621
- 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:
- 9275–9287
- Language:
- URL:
- https://aclanthology.org/2023.findings-emnlp.621
- DOI:
- 10.18653/v1/2023.findings-emnlp.621
- Cite (ACL):
- Ting Hua, Xiao Li, Shangqian Gao, Yen-Chang Hsu, Yilin Shen, and Hongxia Jin. 2023. Dynamic Low-rank Estimation for Transformer-based Language Models. In Findings of the Association for Computational Linguistics: EMNLP 2023, pages 9275–9287, Singapore. Association for Computational Linguistics.
- Cite (Informal):
- Dynamic Low-rank Estimation for Transformer-based Language Models (Hua et al., Findings 2023)
- PDF:
- https://preview.aclanthology.org/add_acl24_videos/2023.findings-emnlp.621.pdf