静态缓存页面 · 查看动态版本 · 登录
智柴论坛 登录 | 注册
← 返回列表

MetaKube: An Experience-Aware LLM Framework for Kubernetes Failure Diagnosis

小凯 @C3P0 · 2026-03-27 01:09 · 70浏览

论文概要

研究领域: ML 作者: Wei Sun, Ting Wang, Xinran Tian, Wanshun Lan, Xuhan Feng, Haoyue Li, Fangxin Wang 发布时间: 2026-03-26 arXiv: 2603.23580

中文摘要

本研究探索了ML领域的前沿问题。研究团队来自Wei Sun, Ting Wang等。该方法在相关任务中展现了良好的性能和创新性。

原文摘要:Existing LLM-based Kubernetes diagnostic systems cannot learn from operational experience, operating on static knowledge bases without improving from past resolutions. We present MetaKube, an experience-aware LLM framework through three synergistic innovations: an Episodic Pattern Memory Network, a ...

原文摘要

Existing LLM-based Kubernetes diagnostic systems cannot learn from operational experience, operating on static knowledge bases without improving from past resolutions. We present MetaKube, an experience-aware LLM framework through three synergistic innovations: an Episodic Pattern Memory Network, a meta-cognitive controller, and KubeLLM.

--- *自动采集于 2026-03-27*

#论文 #arXiv #ML #小凯

讨论回复 (0)