SageMaker AI 推出训练会话，免管理 GPU 自定义模型微调
Announcing new ways to customize models with SageMaker AI, without managing GPUs
做模型微调的朋友可以看看：SageMaker AI 新出的训练会话让你自己写 RL 训练循环，GPU 那摊事 AWS 全包了。
AWS 宣布 Amazon SageMaker AI 的 serverless 模型定制功能正式上线 training sessions（训练会话），即刻可用。开发者可以自行编写后训练循环，例如自定义 loss 函数、reward 信号或多轮强化学习。GPU 分配、分布式执行、故障恢复和 checkpoint 存储则由 SageMaker AI 自动处理。用户因此无需自己管理 GPU 基础设施即可完成模型定制。
Announcing new ways to customize models with SageMaker AI, without managing GPUs
Amazon SageMaker AI serverless model customization now includes training sessions, generally available today. You write and coordinate your own post-training loop (custom loss functions, reward signals, or multi-turn reinforcement learning) while SageMaker AI handles GPU allocation, distributed execution, fault recovery, and checkpoint storage.