Abstract
Conditional masked language model (CMLM) training has proven successful for non-autoregressive and semi-autoregressive sequence generation tasks, such as machine translation. Given a trained CMLM, however, it is not clear what the best inference strategy is. We formulate masked inference as a factorization of conditional probabilities of partial sequences, show that this does not harm performance, and investigate a number of simple heuristics motivated by this perspective. We identify a thresholding strategy that has advantages over the standard “mask-predict” algorithm, and provide analyses of its behavior on machine translation tasks.- Anthology ID:
- 2020.emnlp-main.465
- Volume:
- Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing (EMNLP)
- Month:
- November
- Year:
- 2020
- Address:
- Online
- Editors:
- Bonnie Webber, Trevor Cohn, Yulan He, Yang Liu
- Venue:
- EMNLP
- SIG:
- Publisher:
- Association for Computational Linguistics
- Note:
- Pages:
- 5774–5782
- Language:
- URL:
- https://aclanthology.org/2020.emnlp-main.465
- DOI:
- 10.18653/v1/2020.emnlp-main.465
- Cite (ACL):
- Julia Kreutzer, George Foster, and Colin Cherry. 2020. Inference Strategies for Machine Translation with Conditional Masking. In Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing (EMNLP), pages 5774–5782, Online. Association for Computational Linguistics.
- Cite (Informal):
- Inference Strategies for Machine Translation with Conditional Masking (Kreutzer et al., EMNLP 2020)
- PDF:
- https://preview.aclanthology.org/proper-vol2-ingestion/2020.emnlp-main.465.pdf