github 看到一个项目, 3090 跑 27B, 129tps,最高 207tps
https://github.com/Luce-Org/lucebox-hub DFlash DDtree Qwen3.5 & Qwen3.6 27B GGUF on RTX 3090 First GGUF port of DFlash speculative decod
3090 - 钛刻 - 科技风向旗 - 深度刻画技术趋势,引领数字未来 - 钛刻科技 | TCTI.cn - 钛刻 (TCTI.cn) 为您提供最前沿的硬核科技资讯、深度评测和未来技术趋势分析。
共 13 篇相关文章
https://github.com/Luce-Org/lucebox-hub DFlash DDtree Qwen3.5 & Qwen3.6 27B GGUF on RTX 3090 First GGUF port of DFlash speculative decod
https://github.com/Luce-Org/lucebox-hub DFlash DDtree Qwen3.5 & Qwen3.6 27B GGUF on RTX 3090 First GGUF port of DFlash speculative decod
https://github.com/Luce-Org/lucebox-hub DFlash DDtree Qwen3.5 & Qwen3.6 27B GGUF on RTX 3090 First GGUF port of DFlash speculative decod
https://github.com/Luce-Org/lucebox-hub DFlash DDtree Qwen3.5 & Qwen3.6 27B GGUF on RTX 3090 First GGUF port of DFlash speculative decod
https://github.com/Luce-Org/lucebox-hub DFlash DDtree Qwen3.5 & Qwen3.6 27B GGUF on RTX 3090 First GGUF port of DFlash speculative decod
https://github.com/Luce-Org/lucebox-hub DFlash DDtree Qwen3.5 & Qwen3.6 27B GGUF on RTX 3090 First GGUF port of DFlash speculative decod
https://github.com/Luce-Org/lucebox-hub DFlash DDtree Qwen3.5 & Qwen3.6 27B GGUF on RTX 3090 First GGUF port of DFlash speculative decod
https://github.com/Luce-Org/lucebox-hub DFlash DDtree Qwen3.5 & Qwen3.6 27B GGUF on RTX 3090 First GGUF port of DFlash speculative decod
https://github.com/Luce-Org/lucebox-hub DFlash DDtree Qwen3.5 & Qwen3.6 27B GGUF on RTX 3090 First GGUF port of DFlash speculative decod
https://github.com/Luce-Org/lucebox-hub DFlash DDtree Qwen3.5 & Qwen3.6 27B GGUF on RTX 3090 First GGUF port of DFlash speculative decod
macOS 26.429.30905版本 Codex app 开启后会显示实时思维链 这是个新功能吗? 2 个帖子 - 1 位参与者 阅读完整话题
先说配置: CPU: E5-2680 v4 * 2 内存: 220G GPU: 3090 24G *2 系统:Ubuntu20 工具: LM Studio 0.4.11 Qwen3.6-27B Qwen3.6-35B-A3B 默认专家数量为8,这里佬友自行设置为了9, 问题是:“
两张3090 48G显存 私有化部署35B大模型, 生成速度142.65tokens/s 上下文长度262144 真香 8 个帖子 - 5 位参与者 阅读完整话题