发现信息
- 作者:Yan-Cheng Hsu, Duo Xu, Ziwei Wang
- 年份:2026
- 发表源:Research Square
- OpenAlex ID:https://openalex.org/W7225644782
- DOI:https://doi.org/10.21203/rs.3.rs-11671045/v1
- arXiv:待核验
- 检索主题:code-evolution
- OpenAlex 引用数快照:0(2026-11-12)
摘要线索
Abstract We study risk-aware self-play in finite-action code repair, separating functional success from test-integrity and forbidden-call violations. From 591 MBPP/HumanEval programs, 395 mutation tasks yield 4,527 candidates. On 133 external HumanEval tasks, the full policy reduces expected violations from 68.55% to 0.64% versus public-test-only RL, but fixed multiobjective rewards achieve slightly higher joint success, 72.78% versus 72.37%. Additional experiments execute 5,296 alias/data-flow probes and matched controls, investigate selected-action sparse verification, and test 30 functions from Python, JavaScript and TypeScript repositories. Detector blind spots, identical safe/unsafe feature representations and weak repository transfer limit the safety interpretation. Zero greedy violations on 133 tasks yields only an illustrative 2.23% upper risk bound under iid assumptions. These results support verifier-aware rewards without establishing a consistent self-play advantage or a formal deployment guarantee.
核验与筛选
- 当前状态:待筛选、去重和一手来源核验。
- OpenAlex 元数据和引用数只用于发现,不作为正式引文或质量判断。
- 记录正式版本、venue、方法细节、实验协议和局限后,才迁入稳定论文卡片。