arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2510.09189cs.CL
收藏

LLaMAX2:你的翻译增强模型在推理方面也表现出色

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

National Key Laboratory for Novel Software Technology, Nanjing University(南京大学新型软件技术国家重点实验室);The University of Hong Kong(香港大学);Carnegie Mellon University(卡内基梅隆大学);Shanghai Artificial Intelligent Laboratory(上海人工智能实验室)

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

AI总结:

针对翻译增强模型推理能力弱的问题,提出结合指令微调与层选择性调优的翻译增强方案,推出Qwen3-XPlus,在提升翻译性能的同时保持推理能力,实现多语言增强。

摘要:

通用大型语言模型(LLMs)在推理方面表现出色,但那些为翻译而增强的模型在处理推理任务时却存在困难。为了解决这一问题,我们提出了一种新颖的翻译增强方案,该方案首先对模型进行指令微调,然后仅在平行数据上应用层选择性调优。遵循这一流程,我们推出了Qwen3-XPlus模型,该模型在高资源和低资源语言上的翻译性能均显著提升,在低资源语言(如斯瓦希里语)上实现了15+的spBLEU分数和40+的xComet分数。有趣的是,仅使用小型平行数据集进行训练,Qwen3-XPlus在7项多语言任务上平均提升了1+个点,同时在15个流行的推理数据集上保持了与Qwen3指令模型相当的能力。这项工作为多语言增强提供了一种有前景的方法,显著降低了复杂性,并提高了对更广泛语言的可及性。代码和模型均已公开提供。

英文摘要:

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.

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。

↑