arXiv cs.CL· Shanglin Yang, Zhan Shi·· 11 小时前AI 评分34
研究揭示：短预算 LLM 引导优化中父代继承才是主要增益来源
Lineage Is Memory: Parent Inheritance and Feedback Disclosure in Short-Budget LLM-Guided Optimization
AI 导读
一项覆盖 9 个提示词与程序合成基准、1,750 次预注册实验的研究发现，贪心式父代继承贡献了 LLM 引导优化的大部分进展，即使 seed-reset 重采样每轮获得全部失败的聚类摘要或使用更大的 proposer 模型也无法超越。
正文 · 原文
Abstract:Iterative language-model optimizers refine prompts and programs by evaluating a candidate, showing a proposer model a sample of failing cases, and keeping revisions that improve performance. Recent work attributes the limits of this loop to narrow single-round reflection, introducing full-set error clustering, targeted minibatches, or larger proposer models so the optimizer sees more failure modes per step. Yet every existing evaluation tests these mechanisms exclusively inside lineage-preserving search, where each revision edits a surviving parent, leaving the contribution of parent inheritance confounded with the breadth of per-round feedback. Across 1,750 pre-registered runs spanning nine prompt and program synthesis benchmarks, we separate parent inheritance from feedback disclosure under matched evaluation budgets. We find that greedy parent inheritance drives the vast majority of optimization progress, outperforming seed-reset resampling even when seed-reset proposals receive a clustered summary of all failures in every round or use a larger proposer model. While full-set error clustering and diverse failure selection improve single-step rewrites from the seed and accelerate the opening rounds of search, sequential parent inheritance rapidly closes this gap as solved error modes are retained in the candidate and filtered from subsequent failure lists. Moreover, decomposing token expenditure in GEPA reveals that proposer reflection accounts for only a small fraction of total cost whereas validation and minibatch gating consume nearly the entire budget, showing that greedy parent inheritance over raw failures captures the gains of complex reflective loops at a fraction of their compute.
| Comments: | 17 pages, 2 figures, 10 tables. Code, pre-registrations, and 1,750 run logs at this https URL |
| Subjects: | Software Engineering (cs.SE); Artificial Intelligence (cs.AI); Computation and Language (cs.CL) |
| Cite as: | arXiv:2610.41157 [cs.SE] |
| (or arXiv:2610.41157v1 [cs.SE] for this version) | |
| https://doi.org/10.48550/arXiv.2610.41157 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Zhan Shi [view email]
[v1]
Fri, 30 Oct 2026 02:31:45 UTC (215 KB)
来源：arXiv cs.CL · arxiv.org