arXiv · Computers and Society· Bryan Chen Zhengyu Tan, Weihua Zheng, Thong T. Doan, Bich Ngoc Doan, Jia Wang Peh, Xiaoyuan Yi, Jing Yao, Xing Xie, Nancy F. Chen, Zhengyuan Liu, JinYeong Bak, Wafi Shamdi, Soo Kai Chie, Liew Yu Siong, Aina Azyyati Binti Mohamad Rezal, Lew Yan Yan Vanessa, Huadan Wu, Dylan Raharja, Nadya Yuki Wangsajaya, Akane Fukushige, Kazushi Kato, Koji Inoue, Tatsuya Kawahara, Jaehyung Seo, Dongjun Kim, Seungyoon Lee, Zi Haur Pang, Rui Yang Tan, Charibeth Ko Cheng, Maria Regina Justina Estuar, Jann Railey Montalan, Pham Minh Duc, Roy Ka-Wei Lee·· 7 小时前AI 评分34
CultureConverse:面向东亚与东南亚的多语言多轮文化辅助模拟评测框架
CultureConverse: A Multilingual Multi-turn Simulation Harness for Culturally Grounded Assistance in East and Southeast Asia
AI 导读
CultureConverse 是一个覆盖东亚与东南亚 10 个地区、58 个亚群身份、7 个领域的多语言多轮模拟评测框架,生成 14,610 条评测集与 274,295 条引导对话。在 18 个模型评测中,GPT-5 mini 辅助质量最高;在 27,860 条高质量样本上微调可提升领域内表现并迁移至文化 MCQ 与安全分类基准。
来源:arXiv · Computers and Society · arxiv.org