arXiv cs.CL· Linjuan Wu, Yiwen Qiu, Xinze Lyu, Dingming Li, Hui Yao, Zhongyi Sang, Ye Guo, Daoxin Zhang, Zhe Xu, Fei zhao, Mu Chuan, Yao Hu, Weiming Lu, Yongliang Shen·· 11 小时前AI 评分22
检索增强文化评估助力社交媒体翻译训练：CUE-MT 模型发布
Retrieval-Enhanced Cultural Evaluation for Social Media Translation Training
AI 导读
研究者用检索增强文化评估训练出 CUE-MT 系列社交媒体翻译模型，覆盖 9B、27B 和 35B-A3B 三种骨干规模，在 CUE-MT-Bench 和三个公开基准上优于同规模开源基线。人工评估显示，各规模组文化有效性比最强外部基线提升 7.44-15.12 个百分点；检索使 Judger 四分类准确率比无检索基线提升 8.07 个百分点。
正文 · 原文
Authors:Linjuan Wu, Yiwen Qiu, Xinze Lyu, Dingming Li, Hui Yao, Zhongyi Sang, Ye Guo, Daoxin Zhang, Zhe Xu, Fei zhao, Mu Chuan, Yao Hu, Weiming Lu, Yongliang Shen
Abstract:Social media translation requires preserving meaning while capturing the cultural nuances of user-generated content (UGC). However, gaps in the cultural knowledge of LLM-based evaluators can lead to inaccurate assessments and unreliable training rewards. To improve these signals, we use retrieval-enhanced cultural evaluation to train CUE-MT, a family of social media translation models. A calibrated, domain-specific Judger guides supervised data refinement and is augmented by an Agent for hard-example mining and reinforcement learning. The Agent retrieves relevant external evidence as needed and distills it into concise hints that the Judger uses to score translations and assign rewards. Across 9B, 27B, and 35B-A3B backbones, CUE-MT outperforms the evaluated open-source baselines of comparable size on CUE-MT-Bench and three public benchmarks. Human evaluation on CUE-MT-Bench shows cultural effectiveness gains of 7.44-15.12 percentage points over the strongest external baseline in each size group. Retrieval improves the Judger's four-way classification accuracy by 8.07 percentage points over a matched no-retrieval baseline that uses LLM-generated hints alone. Training ablations further support the combination of hard-example mining and Agent-enhanced rewards.
| Comments: | 17 pages. Preprint |
| Subjects: | Computation and Language (cs.CL) |
| Cite as: | arXiv:2611.02009 [cs.CL] |
| (or arXiv:2611.02009v1 [cs.CL] for this version) | |
| https://doi.org/10.48550/arXiv.2611.02009 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Linjuan Wu [view email]
[v1]
Sun, 1 Nov 2026 15:52:25 UTC (2,449 KB)
来源：arXiv cs.CL · arxiv.org