EMNLP 2021main95 citations
Data and Parameter Scaling Laws for Neural Machine Translation
Mitchell A Gordon, Kevin Duh, Jared Kaplan
Abstract
We observe that the development cross-entropy loss of supervised neural machine translation models scales like a power law with the amount of training data and the number of non-embedding parameters in the model. We discuss some practical implications of these results, such as predicting BLEU achieved by large scale models and predicting the ROI of labeling data in low-resource language pairs.
BibTeX
@inproceedings{gordon-etal-2021-data,
title = "Data and Parameter Scaling Laws for Neural Machine Translation",
author = "Gordon, Mitchell A and
Duh, Kevin and
Kaplan, Jared",
editor = "Moens, Marie-Francine and
Huang, Xuanjing and
Specia, Lucia and
Yih, Scott Wen-tau",
booktitle = "Proceedings of the 2021 Conference on Empirical Methods in Natural Language Processing",
month = nov,
year = "2021",
address = "Online and Punta Cana, Dominican Republic",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2021.emnlp-main.478/",
doi = "10.18653/v1/2021.emnlp-main.478",
pages = "5915--5922"
}