多语言大语言模型的高效路径：通过后训练PARAM$Δ$整合到再利用MoE进行语言扩展
A Data-Efficient Path to Multilingual LLMs: Language Expansion via Post-training PARAM$Δ$ Integration into Upcycled MoE
AI总结：
本文提出了一种高效的方法，通过将密集模型转换为MoE架构，并将不同语言分配给不同专家，从而在不进行复杂对齐阶段的情况下提升多语言大语言模型的性能，同时保留原始能力。
摘要：
将大型语言模型（LLMs）扩展到新语言是一个成本高昂的过程，需要大量的持续预训练（CPT）和数据密集型对齐。尽管最近的数据免费融合技术试图通过将多语言CPT增强模型与其指令版本融合来绕过对齐，但它们受到关键权衡的限制：缓解参数冲突以保持原始能力不可避免地会稀释新语言的学习，反之亦然。为了解决这一矛盾，我们引入了\method，将密集模型重新利用为专家混合（MoE）架构，将不同专家分配给不同语言。然后通过将MoE扩展的参数delta（$Δ_{ ext{post}}$）嫁接回CPT增强的基模型来转移对齐能力，从而绕过复杂的对齐阶段。实验表明，\method在具有相似FLOPs或参数数量的基线方法上表现出色；它在扩展语言上提高了性能，同时有效保留了原始能力。我们进一步证明，我们的方法在不同模型和后训练delta上具有高度适用性。
英文摘要：
Expanding Large Language Models~(LLMs) to new languages is a costly endeavor, demanding extensive Continued Pre-Training~(CPT) and data-intensive alignment. While recent data-free merging techniques attempt to bypass alignment by fusing a multilingual CPT-enhanced model with its instruct counterpart, they are plagued by a critical trade-off: mitigating parameter conflicts to preserve original abilities inevitably dilutes new language acquisition, and vice-versa. To resolve this conflict, we introduce \method, which upcycles a dense model into a Mixture-of-Experts~(MoE) architecture, allocating different experts to different languages. Alignment ability is then transferred by grafting a MoE-expanded parameter delta~($Δ_{\text{post}}$) to the CPT-enhanced base model, bypassing the complex alignment phase. Experiments demonstrate \method's superiority even against baselines with similar FLOPs or number of parameters; it improves performance on expanded languages while effectively preserving original capabilities. We further show our approach is highly applicable across different models and Post-training deltas.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
