AIBullisharXiv – CS AI · Jun 57/10
🧠
Multilingual Fine-Tuning via Localized Gradient Conflict Resolution
Researchers introduce Bucket-Level MOO, a distributed framework that addresses negative interference when fine-tuning Large Language Models across multiple languages by reformulating the problem as multi-objective optimization. The method enables conflict-aware parameter updates without excessive communication overhead while theoretically guaranteeing Refined Pareto Stationarity, improving multilingual performance across four LLM architectures.