AdaRepair-Mem: Adaptive Experience Orchestration for Repository-Level Program Repair

ID: 3391f726
原文链接: https://arxiv.org/abs/2609.20130
PDF: https://arxiv.org/pdf/2609.20130
作者: Z. C. Luo, J. C. Guo, W. J. He, S. Y. Wang, J. C. Yu, F. M. Zhao, Y. Chen, T. Cao, L. Q. Liu, N. Zheng, W. Xu, J. Jiang, Z. M. Zhao
日期: 2026-09-17
更新: 2026-09-17
分类: coding
来源类型: paper
标签: arxiv, paper, program-repair, memory-augmented, swe-bench
质量评分: 4/5
抓取时间: 2026-09-21T04:30Z

中文导读

论文先给出现有仓库级记忆检索的三个结构性问题：episodic memory 在仓库间高度不均衡，低资源仓库拿不到有效支持；记忆越多修复成功率并不单调上升；经验积累与修复阶段错配——可能一堆复现经验、缺 patch/refinement 经验。

AdaRepair-Mem 的三件套：coverage-aware retrieval（同仓记忆不足时回退跨仓库或按修复类型）、quality-aware selection（按相关性 / 历史效用 / 特异性 / 冗余度排序）、stage-aware routing（reproduction / localization / patch / refinement / validation 分阶段路由）。

在 SWE-Bench-Lite 与 SWE-Bench-Verified 上：低覆盖仓库修复率提升、噪声记忆检索减少、fail-to-fixed 的 refinement 转换更好。

结论对做 agent memory 的有直接参考：关键不是攒更多经验，而是“对的修复上下文取回对的记忆”。

为什么值得关注

AdaRepair-Mem 把仓库级修复的记忆检索拆成 coverage / quality / stage 三层，在 SWE-Bench 上拿稳低覆盖仓库修复率

关键信息

论文标题：AdaRepair-Mem: Adaptive Experience Orchestration for Repository-Level Program Repair
作者：Z. C. Luo, J. C. Guo, W. J. He, S. Y. Wang, J. C. Yu, F. M. Zhao, Y. Chen, T. Cao, L. Q. Liu, N. Zheng, W. Xu, J. Jiang, Z. M. Zhao
arXiv：https://arxiv.org/abs/2609.20130
发布时间：2026-09-17
arXiv 分类：cs.SE, cs.AI
关联标签：arxiv, paper, program-repair, memory-augmented, swe-bench

English Abstract

Recent memory-augmented repository-level program repair methods reuse historical repair experiences to improve LLM-based issue resolution. However, our analysis reveals three limitations in existing repository-level memory retrieval. First, episodic memory is highly imbalanced across repositories, leaving low-resource repositories with little effective support. Second, more memory does not monotonically lead to higher repair success, suggesting that relevance, quality, and redundancy matter more than raw memory volume. Third, memory accumulation is phase-misaligned: repositories may contain many reproduction experiences but few patch or refinement experiences. To address these problems, we propose an adaptive experience retrieval framework for repository-level program repair. Our framework introduces coverage-aware retrieval, which falls back to cross-repository or repair-type-based memories when same-repository memory is insufficient; quality-aware selection, which ranks memories by relevance, historical utility, specificity, and redundancy; and stage-aware routing, which separates and retrieves memories for reproduction, localization, patch generation, patch refinement, and validation. Evaluated on SWE-Bench-Lite and SWE-Bench-Verified, the proposed framework improves repair performance on under-covered repositories, reduces noisy memory retrieval, and better supports failed-to-fixed patch refinement. Our results show that the key to memory-augmented repair is not simply accumulating more experiences, but retrieving the right experiences for the right repair context.

English Summary

Obsidian Notes

内容由 opencli arxiv paper 2609.20130 -f json 拉取 arXiv 元数据与摘要生成。
中文导读与价值判断均锚定在条目已有摘要、论文摘要、作者、日期与分类信息上；未补充论文摘要之外的实验细节。