{"ID":2857462,"CreatedAt":"2026-06-01T04:54:23.091178241Z","UpdatedAt":"2026-06-01T04:54:23.091178241Z","DeletedAt":null,"paper_url":"https://arxiv.org/abs/2510.09189","arxiv_id":"2510.09189","title":"LLaMAX2: Your Translation-Enhanced Model also Performs Well in Reasoning","abstract":"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.","short_abstract":"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 Q...","url_abs":"https://arxiv.org/abs/2510.09189","url_pdf":"https://arxiv.org/pdf/2510.09189v1","authors":"[\"Changjiang Gao\",\"Zixian Huang\",\"Jingyang Gong\",\"Shujian Huang\",\"Lei Li\",\"Fei Yuan\"]","published":"2025-10-10T09:33:28Z","proceeding":"cs.CL","tasks":"[\"cs.CL\"]","methods":"[\"Large Language Model\",\"Language Model\"]","has_code":false}