arXiv cs.AI· Pei Yang, Wanyi Chen, Asuka Yuxi Zheng, Xueqian Li, Xiang Li, Haoqin Tu, Jie Xiao, Yifan Pang, Dongdong Zhang, Fuqiang Li, Alfred Long, Lynn Ai, Eric Yang, Bill Shi, Yifan Chen, Shuhan Sun·· 12 小时前AI 评分38
AOI：将失败轨迹转化为训练信号，实现自主云诊断
AOI: Turning Failed Trajectories into Training Signals for Autonomous Cloud Diagnosis
AI 导读
多智能体框架 AOI 在 AIOpsLab 基准的全部 86 项任务上取得 66.3% 的 best@5 成功率，比此前 SOTA 的 41.9% 高出 24.4 个百分点。
正文 · 原文
Authors:Pei Yang, Wanyi Chen, Asuka Yuxi Zheng, Xueqian Li, Xiang Li, Haoqin Tu, Jie Xiao, Yifan Pang, Dongdong Zhang, Fuqiang Li, Alfred Long, Lynn Ai, Eric Yang, Bill Shi, Yifan Chen, Shuhan Sun
Abstract:Large language model (LLM) agents offer a promising data-driven approach to automating Site Reliability Engineering (SRE), yet their enterprise deployment is constrained by three challenges: restricted access to proprietary data, unsafe action execution under permission-governed environments, and the inability of closed systems to improve from failures. We present AOI (Autonomous Operations Intelligence), a trainable multi-agent framework formulating automated operations as a structured trajectory learning problem under security constraints. Our approach integrates three key components. First, a trainable diagnostic system applies Group Relative Policy Optimization (GRPO) to distill expert-level knowledge into locally deployed open-source models, enabling preference-based learning without exposing sensitive data. Second, a read-write separated execution architecture decomposes operational trajectories into observation, reasoning, and action phases, allowing safe learning while preventing unauthorized state mutation. Third, a Failure Trajectory Closed-Loop Evolver mines unsuccessful trajectories and converts them into corrective supervision signals, enabling continual data augmentation. Evaluated on the AIOpsLab benchmark, our contributions yield cumulative gains. (1) The AOI runtime alone achieves 66.3% best@5 success on all 86 tasks, outperforming the prior state-of-the-art (41.9%) by 24.4 points. (2) Adding Observer GRPO training, a locally deployed 14B model reaches 42.9% avg@1 on 63 held-out tasks with unseen fault types, surpassing Claude Sonnet 4.5. (3) The Evolver converts 37 failed trajectories into diagnostic guidance, improving end-to-end avg@5 by 4.8 points while reducing variance by 35%.
| Subjects: | Machine Learning (cs.LG); Artificial Intelligence (cs.AI) |
| Cite as: | arXiv:2603.03378 [cs.LG] |
| (or arXiv:2603.03378v4 [cs.LG] for this version) | |
| https://doi.org/10.48550/arXiv.2603.03378 arXiv-issued DOI via DataCite |
Submission history
From: Yifan Chen [view email]
[v1]
Tue, 3 Mar 2026 02:57:33 UTC (307 KB)
[v2]
Thu, 5 Mar 2026 02:44:23 UTC (307 KB)
[v3]
Tue, 17 Mar 2026 05:26:02 UTC (376 KB)
[v4]
Fri, 16 Oct 2026 03:43:29 UTC (376 KB)
来源：arXiv cs.AI · arxiv.org