Amazon SageMaker AI 多轮强化学习实战指南
原标题：Multi-Turn Reinforcement Learning on Amazon SageMaker AI: A hands-on guide
AI 摘要
AWS 发布技术指南，介绍在 Amazon SageMaker AI 上进行多轮强化学习（MTRL）的完整流程：由 SageMaker AI 负责训练作业与策略模型，AgentCore 托管智能体、环境与奖励函数，通过 GRPO 更新 LoRA 适配器。示例以 Wordle 游戏为任务，将 gpt-oss-20b 在留出词上的求解率从 61% 提升至 84%（默认推理强度），低推理强度下从 15% 提升至 69%，训练 token 成本约 339 美元。文章还说明了支持的基座模型、区域、按 token 计费方式及奖励设计经验。
正文节选
Multi-Turn Reinforcement Learning on Amazon SageMaker AI: A hands-on guide Most fine-tuning teaches a model to produce a better answer to a single prompt. Agentic tasks are different. An agent that files an expense report or triages a support ticket takes a sequence of actions, and each step depends on what came before. There’s rarely one correct sequence to imitate, and the only reliable feedback is whether the whole sequence succeeded. Supervised fine-tuning has no labels for that, and single-