Abstract
Position representation is crucial for building position-aware representations in Transformers. Existing position representations suffer from a lack of generalization to test data with unseen lengths or high computational cost. We investigate shifted absolute position embedding (SHAPE) to address both issues. The basic idea of SHAPE is to achieve shift invariance, which is a key property of recent successful position representations, by randomly shifting absolute positions during training. We demonstrate that SHAPE is empirically comparable to its counterpart while being simpler and faster.- Anthology ID:
- 2021.emnlp-main.266
- Volume:
- Proceedings of the 2021 Conference on Empirical Methods in Natural Language Processing
- Month:
- November
- Year:
- 2021
- Address:
- Online and Punta Cana, Dominican Republic
- Editors:
- Marie-Francine Moens, Xuanjing Huang, Lucia Specia, Scott Wen-tau Yih
- Venue:
- EMNLP
- SIG:
- Publisher:
- Association for Computational Linguistics
- Note:
- Pages:
- 3309–3321
- Language:
- URL:
- https://aclanthology.org/2021.emnlp-main.266
- DOI:
- 10.18653/v1/2021.emnlp-main.266
- Cite (ACL):
- Shun Kiyono, Sosuke Kobayashi, Jun Suzuki, and Kentaro Inui. 2021. SHAPE: Shifted Absolute Position Embedding for Transformers. In Proceedings of the 2021 Conference on Empirical Methods in Natural Language Processing, pages 3309–3321, Online and Punta Cana, Dominican Republic. Association for Computational Linguistics.
- Cite (Informal):
- SHAPE: Shifted Absolute Position Embedding for Transformers (Kiyono et al., EMNLP 2021)
- PDF:
- https://preview.aclanthology.org/nschneid-patch-1/2021.emnlp-main.266.pdf