Simplicity Bias in Transformers and their Ability to Learn Sparse Boolean Functions
Satwik Bhattamishra, Arkil Patel, Varun Kanade, Phil Blunsom
Abstract
Despite the widespread success of Transformers on NLP tasks, recent works have found that they struggle to model several formal languages when compared to recurrent models. This raises the question of why Transformers perform well in practice and whether they have any properties that enable them to generalize better than recurrent models. In this work, we conduct an extensive empirical study on Boolean functions to demonstrate the following: (i) Random Transformers are relatively more biased towards functions of low sensitivity. (ii) When trained on Boolean functions, both Transformers and LSTMs prioritize learning functions of low sensitivity, with Transformers ultimately converging to functions of lower sensitivity. (iii) On sparse Boolean functions which have low sensitivity, we find that Transformers generalize near perfectly even in the presence of noisy labels whereas LSTMs overfit and achieve poor generalization accuracy. Overall, our results provide strong quantifiable evidence that suggests differences in the inductive biases of Transformers and recurrent models which may help explain Transformer’s effective generalization performance despite relatively limited expressiveness.- Anthology ID:
- 2023.acl-long.317
- Volume:
- Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)
- Month:
- July
- Year:
- 2023
- Address:
- Toronto, Canada
- Editors:
- Anna Rogers, Jordan Boyd-Graber, Naoaki Okazaki
- Venue:
- ACL
- SIG:
- Publisher:
- Association for Computational Linguistics
- Note:
- Pages:
- 5767–5791
- Language:
- URL:
- https://aclanthology.org/2023.acl-long.317
- DOI:
- 10.18653/v1/2023.acl-long.317
- Cite (ACL):
- Satwik Bhattamishra, Arkil Patel, Varun Kanade, and Phil Blunsom. 2023. Simplicity Bias in Transformers and their Ability to Learn Sparse Boolean Functions. In Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pages 5767–5791, Toronto, Canada. Association for Computational Linguistics.
- Cite (Informal):
- Simplicity Bias in Transformers and their Ability to Learn Sparse Boolean Functions (Bhattamishra et al., ACL 2023)
- PDF:
- https://preview.aclanthology.org/nschneid-patch-2/2023.acl-long.317.pdf