Admission Without Answers: Label-Free Certification and Experience Learning for LLM-Based Optimization Modeling
无答案准入:基于大语言模型的优化建模的无标注认证与经验学习
机构 * Institute of Operations Research and Analytics(运营研究与分析研究所) ; National University of Singapore(新加坡国立大学) ; Wenzhou Buyi Pharmacy(温州布衣药房) ; College of Computer Science and Artificial Intelligence(计算机科学与人工智能学院) ; Wenzhou University(温州大学)
AI总结 本文提出AdmitOR这一无标注准入机制,通过跨家族团校准阈值实现模型接纳,在优化建模任务中显著提升接纳精度并减少污染接纳,为无标注场景下的模型准入提供了新方案。
Comments Code and data are available at this https URL (https://github.com/junbolian/AdmitOR)