arXiv · cs.CL· Samuel Cahyawijaya, Peerat Limkonchotiwat, Tack Hwa Wong, Hitesh Laxmichand Patel, Amit Agarwal, Manuel Antonio Rufino, Carlos Rafael Catalan, Muhammad Reza Qorib, Vicky Feliren, Holy Lovenia, Aye Hninn Khine, Frederikus Hudi, David Anugraha, Alham Fikri Aji, Romrawin Chumpu, Viet-Thanh Pham, Minghan Wang, Mohamed Fazli Imam, Ruochen Zhang, Joseph Marvin Imperial, Khumaisa Nur'aini, Do Xuan Long, Musa Izzanardi Wijanarko, Joel Ruben Antony Moniz, Patrick Amadeus Irawan, Hanif Muhammad Zhafran, Isaiah Flores, Salsabila Zahirah Pranida, Jun Kevin, Jostin Jerico Rosal, Patricia Nicole Monderin, Kun Kerdthaisong, Ahmad Mustafid, My Chiffon Nguyen, Natchapon Jongwiriyanurak, Siva Worajitwannakul, Haochen Li, Adrian Xuan Wei Lim, Bin Wang, Muhammad Ravi Shulthan Habibi, Lynnette Hui Xian Ng, Mithil Bangera, Yeshil Bangera, Priyaranjan Pattnayak, Dun Li Chan, Sherissa Caren Djuniwar, Cho Chan Myei Oo, Hee Ming Shan·· 17 小时前
多模态视觉语言模型的人类活动区域适配研究：GG-EZ 方法在东南亚实现 5-15% 文化相关性提升
Anthropogenic Regional Adaptation in Multimodal Vision-Language Model
AI 导读
研究者提出"人类活动区域适配"新范式及 GG-EZ 适配方法，通过区域数据过滤与模型合并，在东南亚区域适配中使文化相关性指标提升 5-15%，同时保留超过 98% 的全球性能。实验覆盖大型视觉语言模型、文生图扩散模型和视觉语言嵌入向量模型三类 VL 架构。
来源：arXiv · cs.CL · arxiv.org