English

Low-Rank Adaptation for Multilingual Summarization: An Empirical Study

Computation and Language 2024-04-02 v2 Artificial Intelligence Machine Learning

Abstract

Although the advancements of pre-trained Large Language Models have significantly accelerated recent progress in NLP, their ever-increasing size poses significant challenges for conventional fine-tuning, especially in memory-intensive tasks. We investigate the potential of Parameter-Efficient Fine-Tuning, focusing on Low-Rank Adaptation (LoRA), in the domain of multilingual summarization, a task that is both challenging (due to typically long inputs), and relatively unexplored. We conduct an extensive study across different data availability scenarios, including high- and low-data settings, and cross-lingual transfer, leveraging models of different sizes. Our findings reveal that LoRA is competitive with full fine-tuning when trained with high quantities of data, and excels in low-data scenarios and cross-lingual transfer. We also study different strategies for few-shot cross-lingual transfer, finding that continued LoRA tuning outperforms full fine-tuning and the dynamic composition of language-specific LoRA modules.

Keywords

Cite

@article{arxiv.2311.08572,
  title  = {Low-Rank Adaptation for Multilingual Summarization: An Empirical Study},
  author = {Chenxi Whitehouse and Fantine Huot and Jasmijn Bastings and Mostafa Dehghani and Chu-Cheng Lin and Mirella Lapata},
  journal= {arXiv preprint arXiv:2311.08572},
  year   = {2024}
}

Comments

Findings of NAACL 2024

R2 v1 2026-06-28T13:21:27.342Z