Language and Task Arithmetic with Parameter-Efficient Layers for Zero-Shot Summarization

TL;DR

This paper introduces a method to improve zero-shot cross-lingual summarization by composing parameter-efficient modules for languages and tasks, leveraging unlabeled data and minimal training.

Contribution

It proposes a novel approach to compose language and task-specific PEFT modules arithmetically, enhancing zero-shot transfer across many languages.

Findings

Consistent performance gains in zero-shot summarization across multiple languages.

Effective use of unlabeled data and minimal PEFT training.

Improved transfer by composing modules trained on related languages.

Abstract

Parameter-efficient fine-tuning (PEFT) using labeled task data can significantly improve the performance of large language models (LLMs) on the downstream task. However, there are 7000 languages in the world and many of these languages lack labeled data for real-world language generation tasks. In this paper, we propose to improve zero-shot cross-lingual transfer by composing language or task specialized parameters. Our method composes language and task PEFT modules via element-wise arithmetic operations to leverage unlabeled data and English labeled data. We extend our approach to cases where labeled data from more languages is available and propose to arithmetically compose PEFT modules trained on languages related to the target. Empirical results on summarization demonstrate that our method is an effective strategy that obtains consistent gains using minimal training of PEFT modules.