LLaMAX2：您的翻譯增強模型在推理任務中同樣表現出色

LLaMAX2: Your Translation-Enhanced Model also Performs Well in Reasoning

October 10, 2025

作者: Changjiang Gao, Zixian Huang, Jingyang Gong, Shujian Huang, Lei Li, Fei Yuan

cs.AI

摘要

通用的大型語言模型（LLMs）在推理方面表現出色，但針對翻譯優化的模型在推理任務上卻顯得力不從心。為解決這一問題，我們提出了一種新穎的翻譯增強方案，該方案從指令模型入手，並僅在平行數據上進行層選擇性調優。遵循這一流程，我們推出了Qwen3-XPlus模型，該模型在高資源和低資源語言上的翻譯性能均顯著提升，在低資源語言如斯瓦希里語中，達到了15+的spBLEU和40+的xComet評分。值得注意的是，僅使用小型平行數據集進行訓練，Qwen3-XPlus在7項多語言任務上平均提升了1+分，同時在15個流行的推理數據集上保持了與Qwen3指令模型相當的熟練度。這項工作為多語言增強提供了一種有前景的方法，顯著降低了複雜性，並提高了對更廣泛語言的易用性。代碼和模型均已公開。

English

General Large Language Models (LLMs) excel in reasoning, but those enhanced for translation struggle with reasoning tasks. To address this, we propose a novel translationenhanced recipe that begins with instruct models and applies layer-selective tuning only on parallel data. Following this pipeline, we introduce the Qwen3-XPlus models, which demonstrate significant improvements in translation performance across both high- and lowresource languages, achieving 15+ spBLEU and 40+ xComet in low-resource languages, like Swahili. Interestingly, training only with small parallel datasets, Qwen3-XPlus achieves an average improvement of 1+ points on 7 multilingual tasks while maintaining proficiency comparable to the Qwen3 instruct model in 15 popular reasoning datasets. This work offers a promising approach to multilingual enhancement, significantly reducing complexity and enhancing accessibility for a wider range of languages. The code and model are publicly available.