1. 11
    1. 3

      Why llamafile not llama.cpp?

      1. 4

        Easier for people (including me) to install because there’s only one file to download - it’s basically a convenient wrapper around llama.cpp.

        Also I only wrote about things I’ve got working myself, and I haven’t run Llama 3 with llama.cpp personally yet.

        1. 2

          Useful comment on Mastodon from someone who got llama.cpp to work: https://social.brunk.io/@soeren/112315889818024916

          git clone https://github.com/ggerganov/llama.cpp.git
          make
          huggingface-cli download MaziyarPanahi/Meta-Llama-3-70B-Instruct-GGUF  --include '*Q5_K_M.gguf' --local-dir .
          ./server -m Meta-Llama-3-70B-Instruct.Q5_K_M.gguf -c 8192 -n 2048
          llm -m llamafile "3 neat characteristics of a pelican"