Online Language Adaptive Sampling for Better Distributed Cross-lingual Gains

2026-09-13

AI Summary

为解决低资源语言在跨语言迁移时的对齐问题，提出了一种基于可训练采样概率的自适应采样策略，通过调整每种语言的出现频率来优化模型训练过程。

Abstract

Realignment is a promising approach for improving the cross-lingual transfer ability of multilingual language models, particularly for extremely low-resource languages (LRLs). However, existing realignment methods rely on uniform and random sampling of parallel sentences across languages, which may be suboptimal under limited batch sizes. In practice, models may benefit from seeing certain languages more frequently, especially those that are poorly aligned, and the optimal distribution can evolve throughout training. In this work, we propose a simple yet effective adaptive sampling strategy that assigns trainable sampling probabilities to each language. Languages that contribute more to the realignment loss are sampled more frequently in subsequent batches, and the optimal distribution can evolve throughout training. Our method employs an inner-outer optimization loop with a small overhead, leading to consistent performance improvements and, more importantly, distributing the gains across languages. We observed a $+0.67$ average performance increase on all tasks with XLM-R, and $+0.60$ with Gemma 2 9B compared with uniform realignment. Furthermore, our method is robust across different models. Code available at https://github.com/felixgaschi/multilingual-alignment-and-transfer.

Problem

• Cross-lingual Transfer
• Realignment
• Low-resource Languages
• Adaptive Sampling

Innovation

• Adaptive Sampling
• Realignment Loss
• Inner-Outer Optimization Loop
• Cross-lingual Transfer