Condensing Multilingual Knowledge with Lightweight Language-Specific Modules
Haoran Xu, Weiting Tan, Shuyue Stella Li, Yunmo Chen, Benjamin Van Durme, Philipp Koehn, Kenton Murray
Abstract
Incorporating language-specific (LS) modules or Mixture-of-Experts (MoE) are proven methods to boost performance in multilingual model performance, but the scalability of these approaches to hundreds of languages or experts tends to be hard to manage. We present Language-specific Matrix Synthesis (LMS), a novel method that addresses the issue. LMS utilizes parameter-efficient and lightweight modules, reducing the number of parameters while outperforming existing methods, e.g., +1.73 BLEU over Switch Transformer on OPUS-100 multilingual translation. Additionally, we introduce Fuse Distillation (FD) to condense multilingual knowledge from multiple LS modules into a single shared module, improving model inference and storage efficiency. Our approach demonstrates superior scalability and performance compared to state-of-the-art methods.
BibTeX
@inproceedings{
xu2023condensing,
title={Condensing Multilingual Knowledge with Lightweight Language-Specific Modules},
author={Haoran Xu and Weiting Tan and Shuyue Stella Li and Yunmo Chen and Benjamin Van Durme and Philipp Koehn and Kenton Murray},
booktitle={The 2023 Conference on Empirical Methods in Natural Language Processing},
year={2023},
url={https://openreview.net/forum?id=siiVduxdRz}
}