ArchOR：面向优化建模的结构感知技能检索与仲裁方法
Structure-Aware Skill Retrieval and Arbitration for Optimization Modeling
有人在优化建模任务上给 LLM 智能体换了个更聪明的技能检索方式，Pass@1 提到 65%，比 OptSkills 高了快 3 个点，做运筹方向可以看看。
论文提出 ArchOR，一个以原型为中心的技能智能体，在 OptSkills 基础上新增 CARL 和 S³R 两个检索模块：CARL 通过低秩原型残差优化缓存技能表示的排序，S³R 为技能附加开放维度结构签名并按加权 Jaccard 重叠筛选候选。配套的仲裁规则决定结构检索路由何时替代原有选择器。在相同 DeepSeek-V4.1-Flash 骨干和已发布技能库下，ArchOR 在五个基准上取得 65.00% Micro-Avg. Pass@1，较 OptSkills 提升 1.64 个百分点 Micro-Avg. 和 2.68 个百分点 Macro-Avg.。
Structure-Aware Skill Retrieval and Arbitration for Optimization Modeling
Large Language Models (LLMs) can turn natural-language optimization tasks into solver-checked answers, but current agents remain brittle under narrative rewrites and often store reusable experience at the instance level. We introduce \textsc{ArchOR}, an archetype-centric skill agent that builds on \textsc{OptSkills} by adding two retrieval-focused modules. \emph{Centroid-Aware Representation Learning} (CARL) refines the ordering of cached skill representations through a low-rank prototype residual. \emph{Signature-Guided Retrieval} (S\textsuperscript{3}R) attaches open-dimensional structural signatures to skills, prioritizes candidates by weighted Jaccard overlap, and bounds semantic re-ranking to a top-$K$ set. A baseline-preserving arbitration rule determines when the structural route should replace the original selector. With the same DeepSeek-V4.1-Flash backbone and released skill library, \textsc{ArchOR} reaches 65.00\% Micro-Avg.\ Pass@1 across five benchmarks, improving over \textsc{OptSkills} by 1.64~pp Micro-Avg.\ and 2.68~pp Macro-Avg.