RustingStudio/RustingStrata
在 GitHub 打开 ↗Pure-Rust MoE inference engine for Qwen3.5/3.6 GGUF. CPU+GPU hybrid experts, NVRTC CUDA kernels, OpenAI-compatible server. 2x llama.cpp prefill on a 12 GB RTX 3060.
Star43
24h 新增+6
7 天新增—
Fork7
Issue2
语言Rust
项目简介
Pure-Rust MoE inference engine for Qwen3.5/3.6 GGUF. CPU+GPU hybrid experts, NVRTC CUDA kernels, OpenAI-compatible server. 2x llama.cpp prefill on a 12 GB RTX 3060.
Star 趋势
话题标签
cudaggufinferencellama-cppllmlocal-llmmixture-of-expertsmoeopenai-apiqwenrust
仓库信息
| License | Other |
| 创建时间 | 2026-10-10 |
| 最后推送 | 11 小时前 |
| 热度分 | 23.0 |