跳转至

You can run in a variety of ways, I went with llama.cpp,

If my notes are correct:

Clone and build

git clone https://github.com/ggerganov/llama.cpp cd llama.cpp mkdir build && cd build cmake .. -DLLAMA_AVX2=ON -DLLAMA_F16C=ON -DLLAMA_FMA=ON cmake --build . --config Release

原链接