AWS SageMaker AI 推出免管理 GPU 的模型定制训练会话
原标题：Announcing new ways to customize models with SageMaker AI, without managing GPUs
AI 摘要
AWS 宣布 Amazon SageMaker AI 无服务器模型定制功能正式推出 training sessions，用户可自行编写并协调训练循环（含自定义损失函数与奖励信号），而 GPU 分配、分布式执行、故障恢复和检查点存储由 SageMaker AI 托管。该功能支持 SFT、DPO、RLVR、RLAIF、多轮 RL 及 LoRA/全量微调，并提供 cookbook 与模型定制 agent 技能，可在 SageMaker Studio、Kiro、Claude Code、Codex 或 GitHub Copilot 中使用。客户 ImagineSoftware 与 Karini AI 已用其构建医疗理赔分诊模型和基于 Neptune 知识图谱的多轮工具调用微调。
正文节选
Announcing new ways to customize models with SageMaker AI, without managing GPUs Companies getting real results from AI aren’t only using foundation models (FMs) off the shelf. They’re customizing open-weight models using post-training techniques such as reinforcement learning. A model trained on your business context, product catalog, or clinical vocabulary will be more accurate for your task and cost less to run. When you build agents, your data, workflows, evaluations, and success criteria al