Agent Substrate:为智能体扩展 Kubernetes

AI 智能体大部分时间都处于空闲状态,等待人类提示、工具返回结果或模型响应。Kubernetes 并非为此而生。每个会话跑一个 Pod,会为那些什么也没做的会话占着内存;而在多轮之间删除 Pod,每次都要付出数秒的冷启动代价,还会丢掉智能体的状态。RL 后训练也撞上了同一堵墙。Agent Substrate 现已捐赠给……
译文由英文原文机器翻译,术语按行业惯例统一,可在文末展开原文核对。
查看英文原文(Agent Substrate: Advancing Kubernetes for Agents at Scale)
AI agents spend most of their time idle, waiting on a human prompt, a tool result or a model response. Kubernetes wasn’t built for that. Running one Pod per...
AI agents spend most of their time idle, waiting on a human prompt, a tool result or a model response. Kubernetes wasn’t built for that. Running one Pod per session holds memory for sessions that aren’t doing anything, and deleting Pods between turns pays a multi-second cold start every time and loses the agent’s state. RL post-training hits the same wall. Agent Substrate, now donated to the…
正文保留英文原文,标题与摘要由机器翻译。
本文由采集器自动同步
阅读原文 ↗
原文发布于 NVIDIA Developer(Agent Substrate: Advancing Kubernetes for Agents at Scale),标题、摘要与正文由机器翻译,术语按行业惯例统一,未经人工逐句校订。