AdaRepair-Mem: Adaptive Experience Orchestration for Repository-Level Program Repair
AdaRepair-Mem introduces coverage-, quality-, and stage-aware memory retrieval that improves LLM-based repository-level program repair on SWE-Bench.
The paper identifies three limitations in existing repository-level memory retrieval for LLM-based issue resolution: imbalanced episodic memory across repositories, non-monotonic gains from memory volume, and phase-misaligned memory accumulation. The proposed framework adds coverage-aware retrieval with cross-repository fallback, quality-aware selection ranking relevance, utility, specificity, and redundancy, and stage-aware routing across reproduction, localization, patch generation, refinement, and validation. On SWE-Bench-Lite and SWE-Bench-Verified it improves repair of under-covered repositories and reduces noisy retrieval.
- Coverage-aware retrieval falls back to cross-repository or repair-type memories
- Quality-aware selection ranks memories by relevance, utility, specificity, redundancy
- Stage-aware routing separates memory per repair phase
- Evaluated on SWE-Bench-Lite and SWE-Bench-Verified
Full article192 words · extracted from arxiv.org · click to collapse
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.
Text extracted automatically; images, tables and formatting may be missing. Original: https://arxiv.org/abs/2609.20130