Abstract
The Transformer architecture has become prominent in developing large causal language models. However, mechanisms to explain its capabilities are not well understood. Focused on the training process, here we establish a meta-learning view of the Transformer architecture when trained for the causal language modeling task, by explicating an inner optimization process that may happen within the Transformer. Further, from within the inner optimization, we discover and theoretically analyze a special characteristic of the norms of learned token representations within Transformer-based causal language models. Our analysis is supported by experiments conducted on pre-trained large language models and real-world data.- Anthology ID:
- 2024.findings-acl.922
- Volume:
- Findings of the Association for Computational Linguistics ACL 2024
- Month:
- August
- Year:
- 2024
- Address:
- Bangkok, Thailand and virtual meeting
- Editors:
- Lun-Wei Ku, Andre Martins, Vivek Srikumar
- Venue:
- Findings
- SIG:
- Publisher:
- Association for Computational Linguistics
- Note:
- Pages:
- 15612–15622
- Language:
- URL:
- https://aclanthology.org/2024.findings-acl.922
- DOI:
- Cite (ACL):
- Xinbo Wu and Lav Varshney. 2024. A Meta-Learning Perspective on Transformers for Causal Language Modeling. In Findings of the Association for Computational Linguistics ACL 2024, pages 15612–15622, Bangkok, Thailand and virtual meeting. Association for Computational Linguistics.
- Cite (Informal):
- A Meta-Learning Perspective on Transformers for Causal Language Modeling (Wu & Varshney, Findings 2024)
- PDF:
- https://preview.aclanthology.org/nschneid-patch-4/2024.findings-acl.922.pdf