下载SakuraLLM模型鉴于显存为6G，下载20241012-Qwen2.5-1.5B-v1.0模型，去https://hf-mirror.com/SakuraLLM/Sakura-1.5B-Qwen2.5-v1.0-GGUF/tree/main下载gguf文件编译llama.cpp下载llama.cpp代码包cmake-Bbuild-DGGML_CUDA=ONcmake--buildbuild--configRelease将build