Do you host your own ML / AI / LLM? What do you use, and what do you use it for?

  • brucethemoose@lemmy.world
    link
    fedilink
    English
    arrow-up
    2
    ·
    16 hours ago

    CPU offloading is too slow unless you use a hybrid MoE model, with the --n-cpu-moe parameter, specifically.

    This only offloads “sparse” parts of the model to the CPU, which take up a lot of RAM but are very compute-lite to run. In practice, thats most of the size of modern MoE LLMs.

    • robber@lemmy.ml
      link
      fedilink
      English
      arrow-up
      1
      ·
      5 hours ago

      Since implementation of the --fit parameter and its relatives, and --fit on becoming the default, llama.cpp intelligently decides what to offload. For me, it made --n-cpu-moe obsolete.

      • brucethemoose@lemmy.world
        link
        fedilink
        English
        arrow-up
        1
        ·
        4 hours ago

        Mostly, yeah.

        Sometimes it’s better to “cut it close,” with (for instance) a 27B model that’s nearly OOMing your VRAM fully offloaded, but you know will be fine in regular use without too many programs open.

        In my case, with MiMo 2.5, it fills both my CPU and GPU RAM rather completely, so it’s best to set a static value so I don’t swap CPU RAM, and don’t OOM on the GPU either.