diff --git a/content/post/3c1779ab6468800eac67f555a5301e82/index.zh-cn.md b/content/post/3c1779ab6468800eac67f555a5301e82/index.zh-cn.md index 5d8e2ea..dfc49ae 100644 --- a/content/post/3c1779ab6468800eac67f555a5301e82/index.zh-cn.md +++ b/content/post/3c1779ab6468800eac67f555a5301e82/index.zh-cn.md @@ -3,7 +3,7 @@ author = "FlintyLemming" title = "Qwen3.8-27B 部署技术报告" slug = "3c1779ab6468800eac67f555a5301e82" date = "2026-08-19" -description = "A6000 上 FP8 走 Marlin W8A16;4 卡最快 109 tok/s,2 卡驻留用 DFlash2" +description = "最终驻留配置为 FP8 + TP2 + DFlash2,单流 96 tok/s" categories = ["AI"] tags = ["Qwen", "SGLang"] image = "https://assets.flinty.moe/blog/posts/2026/08/Qwen3%208-27B%20%E9%83%A8%E7%BD%B2%E6%8A%80%E6%9C%AF%E6%8A%A5%E5%91%8A/pawel-czerwinski-KMICAjAH7fU-unsplash.jpg"