DeepSpeed
软件
训练
智能体
Microsoft's scaling toolkit — ZeRO sharding made huge models trainable.
您正在查看一条软件记录：AI 技术栈中的一款工具或平台——谁维护它、它运行在什么硬件上。
DeepSpeed's ZeRO optimizer sharding let models far larger than one GPU's memory train efficiently, powering the Megatron-Turing era and countless open efforts since. Its ideas (now echoed by PyTorch FSDP) defined how the industry splits models across thousands of devices.
来源
记录审阅截至 2026-01-15 — 查看 方法论.