摘要：编程 Agent 可以递归地修改自身实现，从而形成自我改进的循环。尽管先前研究表明，这种方式能够提升编程基准测试中的表现，但现有方法成本高昂且计算密集。我们提出了一种简单、样本效率高的自我改进框架，可在严格的预算约束下显著提升编程性能。我们发现，对候选自我修改进行评估是主要的运行时瓶颈：此前方法通常需要使用修改后的 Agent 重新运行一部分基准任务，以估计修改的有效性，而这一过程非常耗时。我们提出通过快速树搜索进行递归自我改进（Recursive Self Improvement via Fast Tree-search，SIFT）：该方法利用 LLM-as-a-judge 信号补充下游任务评估，对候选补丁进行两两比较；随后通过正则化 Bradley-Terry 模型汇总胜负记录，并利用所得强度分数，在轻量级解耦树搜索中驱动基于排名的父节点采样。昂贵的下游任务评估仅保留给最有前景的节点。借助完全解耦的树搜索流水线，评审分数提供中间信号，引导系统探索有潜力的候选补丁，而不会受缓慢评估运行的瓶颈限制。SIFT 在完整 Polyglot 基准上优于现有的基于树搜索的自我演化框架，同时在 CPU 小时数、实际耗时和 API 成本方面具有显著更低的资源需求。

Abstract:Coding agents can recursively modify their own implementations, forming a loop of self-improvement. While prior work shows this can boost performance on coding benchmarks, existing approaches are costly and compute-intensive. We introduce a simple, sample-efficient self-improvement framework that significantly improves coding performance under strict budget constraints. We identify evaluation of candidate self-modifications as the main runtime bottleneck since prior approaches estimate their effectiveness by re-running a subset of benchmark tasks with the modified agent, which is time-consuming. We introduce Recursive Self Improvement via Fast Tree-search (SIFT), which augments these downstream task evaluations with an LLM-as-a-judge signal that performs pairwise comparisons between candidate patches, where the win-loss record is aggregated with a regularized Bradley-Terry model, and the resulting strength scores drive rank-based parent sampling inside a lightweight disaggregated tree search. Expensive downstream task evaluations are reserved only for the most promising nodes. Using a fully disaggregated tree search pipeline, the judge scores provide intermediate signal to guide exploration on promising candidate patches without being bottlenecked by slow evaluation runs. SIFT outperforms existing tree-search based self-evolution frameworks on the full Polyglot benchmark with significantly lower resource requirements in terms of CPU hours, wall clock time, and API cost.