arXiv cs.AI· Jingxuan Xu, Zihang Xu, Yuechen Luo, Ziqi Miao, Wenqiang Zhu, Weihao Li, Junhao Fan, Xinshun Feng, Zhiyi Lai, Yanan Wu, Haoyang Huang, Ken Deng·· 4 小时前AI 评分32
RIPPLE:通过过程级策略学习与效果感知信用实现长程智能体编程的恢复内化
RIPPLE: Recovery Internalization through Process-Level Policy Learning and Effect-Aware Credit for Long-Horizon Agentic Coding
AI 导读
RIPPLE 是一个面向长程编程智能体的恢复导向训练框架,将过程级指导内化为单一自恢复策略,在四个仓库级软件工程基准上平均提升基础策略 4.8 分,部署时无需辅助 critic。
正文 · 原文
Authors:Jingxuan Xu, Zihang Xu, Yuechen Luo, Ziqi Miao, Wenqiang Zhu, Weihao Li, Junhao Fan, Xinshun Feng, Zhiyi Lai, Yanan Wu, Haoyang Huang, Ken Deng
Abstract:Long-horizon coding agents can observe a failure and generate a plausible critique yet continue along the same incorrect trajectory. We call this the Critique-to-Action Handoff Gap. We introduce RIPPLE, a recovery-oriented training framework that internalizes process-level guidance into a single self-recovering policy. A prefix-only teacher provides evidence-grounded recovery supervision without access to future actions or terminal outcomes. After lightweight distillation, reinforcement learning assigns stage-specific credit to reflection triggering, diagnosis, and corrective action. Local validity signals gate trajectory advantages so that intermediate recovery decisions do not indiscriminately inherit terminal outcomes. Across four repository-level software-engineering benchmarks, RIPPLE improves the base policy by an arithmetic mean of 4.8 score points, with no auxiliary critic at deployment. Controlled ablations isolate the benefits of local, stage-aligned credit, and matched-prefix interventions show that critique semantics affect subsequent correction. Recovery gains concentrate where the prefix contains sufficient diagnostic evidence and the repository state remains reversible within the remaining budget.
| Subjects: | Artificial Intelligence (cs.AI) |
| Cite as: | arXiv:2610.31276 [cs.AI] |
| (or arXiv:2610.31276v2 [cs.AI] for this version) | |
| https://doi.org/10.48550/arXiv.2610.31276 arXiv-issued DOI via DataCite |
Submission history
From: Miao Ziqi [view email]
[v1]
Thu, 22 Oct 2026 13:45:33 UTC (5,439 KB)
[v2]
Tue, 17 Nov 2026 09:53:25 UTC (5,441 KB)
来源:arXiv cs.AI · arxiv.org