Exowatt 借助 SageMaker 无服务器微调，以更低成本超越前沿模型准确率
原标题：Exowatt beats frontier models on accuracy with SageMaker serverless model customization
AI 摘要
Exowatt 利用 Amazon SageMaker AI 的无服务器模型定制能力，构建了一条可重复的 AI 流水线，为边缘告警分诊和云端能源市场分析两个场景分别微调 Qwen3-1.7B 与 Qwen3-14B 模型。微调后的专用小模型在特定任务上准确率超过前沿模型，成本最高降低 49 倍。该方案解决了边缘无 GPU、网络不稳定以及大规模部署成本过高的问题，使 Exowatt 能在数天内而非数月内完成从原始数据到可部署模型的流程。
正文节选
Exowatt beats frontier models on accuracy with SageMaker serverless model customization Exowatt builds modular, dispatchable solar systems that deliver reliable power for energy-intensive applications such as AI data centers. Its flagship product, the Exowatt P3, captures solar energy through optical lenses, stores it as heat in a long-duration thermal battery, and dispatches electricity on demand through a heat engine. Unlike grid-tied photovoltaic solar systems, P3 powers data centers directly