X.PIN· @thexpin · X·· 2 小时前AI 评分73
AI 导读
据科创板日报消息人士，DeepSeek 计划在华为昇腾芯片上预训练下一代 V5 模型。此前昇腾只承担 DeepSeek 的推理和 V4-Pro 后训练；作者认为 DeepSeek 于 9 月 30 日开源其完整昇腾算力与通信栈（DeepGEMM、FlashMLA、DeepEP）正与此相关。
正文 · 原文
DeepSeek plans to pre-train its next-gen V5 model on Huawei Ascend chips, per STAR Market Daily sources.
Until now, Ascend only handled DeepSeek inference and V4-Pro post-training. Suddenly, DeepSeek open-sourcing its full Ascend compute & comms stack (DeepGEMM, FlashMLA, DeepEP) on Sept 30 makes total sense.
来源：X.PIN · x.com