在 SageMaker HyperPod 上用 SkyRL 加速多模态 RL 训练
原标题：Accelerate multimodal RL training with SkyRL on Amazon SageMaker HyperPod
AI 摘要
AWS 博客介绍在 Amazon SageMaker HyperPod 上使用开源 RL 框架 SkyRL 进行多模态强化学习训练。方案以 Qwen3-VL-8B 视觉语言模型为策略模型，采用 GRPO 算法训练其走 2D 迷宫，从 VisGym SFT 检查点出发，在固定 64 迷宫评测集上把解迷宫成功率从 43.75% 提升到 95% 以上。HyperPod 提供多节点集群韧性、自动替换故障节点、检查点恢复和 Grafana 监控，支撑长时间多节点 RL rollout。
正文节选
Accelerate multimodal RL training with SkyRL on Amazon SageMaker HyperPod Reinforcement learning (RL) post-training is becoming a standard step in building capable language model agents. Models learn to reason and act across sequences of steps by generating trajectories, receiving rewards, and updating their policy based on outcomes. Running this at scale, across multiple nodes with hundreds of GPU-hours of rollouts per training run, requires persistent cluster infrastructure. That infrastructur