AWS 推出 TorchNeuron：无需改代码即可在 Trainium 上运行 PyTorch 模型
原标题：AWS Launches TorchNeuron to Run PyTorch Models on AWS Trainium Without Modifications
AI 摘要
在 PyTorch Conference North America 2026 上，亚马逊发布原生 PyTorch 后端 TorchNeuron，开发者几乎无需修改代码即可在 AWS Trainium 芯片上运行现有 PyTorch 模型，通常只需将 .to('cuda') 改为 .to('neuron')。该后端原生支持 DDP、FSDP2、DTensor 等分布式训练与并行 API，并允许企业在训练与推理负载间共享和调配硬件资源。亚马逊称已用 10 万个热门模型和 800 多种 ML 架构进行验证，Cartesia 和 Lieber Institute 等机构已采用。此举紧随谷歌推出 TorchTPU 之后，反映云厂商围绕 PyTorch 生态降低自研芯片迁移门槛的竞争。
正文节选
At the recent PyTorch Conference North America 2026, Amazon introduced TorchNeuron, a native PyTorch backend that enables developers to run existing PyTorch code on AWS Trainium AI chips with virtually no code modifications. Beyond reducing engineering effort, TorchNeuron allows enterprises to share hardware capacity between training and inference workloads, reallocating resources from inference to training during off-peak hours. Amazon says that a single configuration change is sufficient to sw