新发现
marco-lu2508/fllm — 为 Q2_K·Q3_K·IQ4_XS 内核添加预取
完全用 Rust 编写的本地 LLM 推理引擎 fLLM,为 Q2_K·Q3_K·IQ4_XS 量化格式的共用内核添加了内存预取。
完全用 Rust 编写的本地 LLM 推理引擎 fLLM,为 Q2_K·Q3_K·IQ4_XS 量化格式的共用内核添加了内存预取。
GitHub 项目说明: Local LLM inference engine in pure Rust: GGUF, OpenAI/Anthropic/Ollama-compatible API, CUDA (incl. NVFP4), Metal, Vulkan, CPU
在 GitHub 查看 ↗完全用 Rust 编写的本地 LLM 推理引擎 fLLM,为 Q2_K·Q3_K·IQ4_XS 量化格式的共用内核添加了内存预取。
没有符合条件的动态。