发表机构
AXXX; Lomonosov Moscow State University; Applied AI Institute（AXXX; 莫斯科罗蒙诺索夫国立大学; 应用人工智能研究院）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
EvoMem是用于LLM驱动的进化程序搜索的持久记忆架构，可捕获复用变异知识，在多任务基准测试中提升了目标指标或搜索速度，减少冗余探索。
AI 中文摘要
在进化代码搜索中，成功的变异策略可能包含可复用的知识，这些知识不仅在单次运行中有用，在某些情况下还可跨相关任务和领域迁移。然而，现有的大语言模型（LLM）驱动的进化框架大多会丢弃这类知识，反复重新发现类似思路，限制了跨运行和跨任务学习的机会。我们提出EvoMem，一种用于LLM驱动的进化程序搜索的持久记忆架构，可捕获并复用候选变异知识。EvoMem将成功的变异事件转换为结构化的、感知任务的建议，供后续运行使用。它分两个阶段运行：每次运行后，提取并存储带有来源的有价值思路；在后续进化过程中，基于当前任务和程序上下文检索一小部分相关指令，以指导变异。在几何优化、多跳问答、GPU内核优化及相关基准测试中，我们的实验显示，在大多数评估设置下，目标指标或搜索速度均有平均正向提升，同时也揭示了任务间的差异性。总体而言，EvoMem提供证据表明，持久记忆可减少部分冗余探索，并提升LLM驱动的进化搜索中成功策略的复用与适配能力。
英文摘要
Successful mutation strategies in evolutionary code search may contain reusable knowledge that is useful beyond a single run, and in some cases may transfer across related tasks and domains. However, existing LLM-driven evolutionary frameworks largely discard such knowledge, repeatedly rediscovering similar ideas and limiting opportunities for cross-run and cross-task learning. We introduce EvoMem, a persistent memory architecture for LLM-based evolutionary program search that captures and reuses candidate mutation knowledge. EvoMem converts successful mutation events into structured, task-aware advice for future runs. It operates in two phases: after each run, it extracts and stores promising ideas with provenance, and during subsequent evolution, it retrieves a small set of relevant instructions based on the current task and program context to guide mutation. Across geometric optimization, multi-hop question answering, GPU kernel optimization, and related benchmarks, our experiments show positive average improvements in target metrics or search speed for most evaluated settings, while also revealing variability across tasks. Overall, EvoMem provides evidence that persistent memory can reduce some redundant exploration and improve the reuse and adaptation of successful strategies in LLM-driven evolutionary search.
