BunshinRL：拆分实时智能体工作区，降低强化学习内存开销
BunshinRL: Splitting Live Agent Workspaces for Memory-Efficient Agentic RL
做智能体强化学习训练的朋友可以看看这篇论文。它用容器级工作区分叉加工具缓存，实测峰值内存砍掉 43%、延迟降 23%，实现细节在 Linux 内核模块层面都讲清楚了，适合自己搭 rollout 基础设施的工程团队参考。
BunshinRL 是一个面向智能体强化学习（RL）的 rollout 系统，针对多环境并发时内存占用高、慢速工具调用拖慢整批 rollout 两大问题提出解决方案。
系统通过三项核心技术实现优化：跨环境内存共享、工具前缀缓存和推测性工具执行。这些技术建立在 BunshinFork 之上——一个针对容器的实时整工作区分叉机制，能够重建进程树、以写时复制方式共享内存，并保留文件、通信和 GUI 状态，支持缓存状态复用与错误推测后的回滚。实现层面涉及 Linux 内核模块和 CRIU 的修改。
实测结果显示，BunshinRL 将峰值内存降低最高 43%，总内存消耗降低最高 51%，端到端 rollout 延迟降低 23%。对需要大规模跑智能体 RL 训练的团队来说，这套系统级优化直接关系到训练成本和吞吐。
BunshinRL: Splitting Live Agent Workspaces for Memory-Efficient Agentic RL
Agentic reinforcement learning (RL) rollouts interleave model reasoning with tool calls in persistent workspaces. Scaling these rollouts incurs substantial memory costs for concurrent environments, while slow tool calls delay entire rollout batches. We present BunshinRL, an agentic RL rollout system that exploits shared environment state and recurring tool calls through three techniques: memory sharing across environments, tool-prefix caching, and speculative tool execution. These techniques build on BunshinFork, a novel live whole-workspace forking mechanism for containers. BunshinFork reconstructs process trees, shares memory copy-on-write, and preserves file, communication, and GUI state, enabling cached-state reuse and rollback after incorrect speculation. We implement BunshinRL and BunshinFork through Linux kernel module and CRIU modifications. BunshinRL reduces peak memory by up to 43%, total memory consumption by up to 51%, and end-to-end rollout latency by 23%.
本条情报归属 AITOP「论文」核心追踪栏目。在跨区域产业观察坐标中，该动态直接关联「AI 开发者工具链、Coding Agent 与工程化实践」研究主线，反映出当前技术演进与商业落地的最新风向。涉及核心实体与技术议题包括：BunshinRL、智能体、强化学习、CRIU，读者可通过下方标签进入对应专题页查阅历史脉络与关联对比。
信源核验说明：本文原始线索采编自「arXiv cs.AI」（@Yutong Huang, Vikranth Srivatsa, Alex Asch, Yiying Zhang），经 AITOP 多维评分管线结构化提炼。