MixCE: Training Autoregressive Language Models by Mixing Forward and Reverse Cross-Entropies
Shiyue Zhang, Shijie Wu, Ozan Irsoy, Steven Lu, Mohit Bansal, Mark Dredze, David Rosenberg
Abstract
Autoregressive language models are trained by minimizing the cross-entropy of the model distribution Q relative to the data distribution P – that is, minimizing the forward cross-entropy, which is equivalent to maximum likelihood estimation (MLE). We have observed that models trained in this way may “over-generalize”, in the sense that they produce non-human-like text. Moreover, we believe that reverse cross-entropy, i.e., the cross-entropy of P relative to Q, is a better reflection of how a human would evaluate text generated by a model. Hence, we propose learning with MixCE, an objective that mixes the forward and reverse cross-entropies. We evaluate models trained with this objective on synthetic data settings (where P is known) and real data, and show that the resulting models yield better generated text without complex decoding strategies.- Anthology ID:
- 2023.acl-long.502
- 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:
- 9027–9050
- Language:
- URL:
- https://preview.aclanthology.org/add_missing_videos/2023.acl-long.502/
- DOI:
- 10.18653/v1/2023.acl-long.502
- Cite (ACL):
- Shiyue Zhang, Shijie Wu, Ozan Irsoy, Steven Lu, Mohit Bansal, Mark Dredze, and David Rosenberg. 2023. MixCE: Training Autoregressive Language Models by Mixing Forward and Reverse Cross-Entropies. In Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pages 9027–9050, Toronto, Canada. Association for Computational Linguistics.
- Cite (Informal):
- MixCE: Training Autoregressive Language Models by Mixing Forward and Reverse Cross-Entropies (Zhang et al., ACL 2023)
- PDF:
- https://preview.aclanthology.org/add_missing_videos/2023.acl-long.502.pdf