Are Pretrained Convolutions Better than Pretrained Transformers?
Yi Tay, Mostafa Dehghani, Jai Prakash Gupta, Vamsi Aribandi, Dara Bahri, Zhen Qin, Donald Metzler
Abstract
In the era of pre-trained language models, Transformers are the de facto choice of model architectures. While recent research has shown promise in entirely convolutional, or CNN, architectures, they have not been explored using the pre-train-fine-tune paradigm. In the context of language models, are convolutional models competitive to Transformers when pre-trained? This paper investigates this research question and presents several interesting findings. Across an extensive set of experiments on 8 datasets/tasks, we find that CNN-based pre-trained models are competitive and outperform their Transformer counterpart in certain scenarios, albeit with caveats. Overall, the findings outlined in this paper suggest that conflating pre-training and architectural advances is misguided and that both advances should be considered independently. We believe our research paves the way for a healthy amount of optimism in alternative architectures.- Anthology ID:
- 2021.acl-long.335
- Volume:
- Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing (Volume 1: Long Papers)
- Month:
- August
- Year:
- 2021
- Address:
- Online
- Editors:
- Chengqing Zong, Fei Xia, Wenjie Li, Roberto Navigli
- Venues:
- ACL | IJCNLP
- SIG:
- Publisher:
- Association for Computational Linguistics
- Note:
- Pages:
- 4349–4359
- Language:
- URL:
- https://aclanthology.org/2021.acl-long.335
- DOI:
- 10.18653/v1/2021.acl-long.335
- Cite (ACL):
- Yi Tay, Mostafa Dehghani, Jai Prakash Gupta, Vamsi Aribandi, Dara Bahri, Zhen Qin, and Donald Metzler. 2021. Are Pretrained Convolutions Better than Pretrained Transformers?. In Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing (Volume 1: Long Papers), pages 4349–4359, Online. Association for Computational Linguistics.
- Cite (Informal):
- Are Pretrained Convolutions Better than Pretrained Transformers? (Tay et al., ACL-IJCNLP 2021)
- PDF:
- https://preview.aclanthology.org/nschneid-patch-2/2021.acl-long.335.pdf
- Code
- google-research/google-research
- Data
- AG News, C4, Civil Comments, IMDb Movie Reviews, MultiNLI, SQuAD, SST, SST-2