• Septimaeus@infosec.pub
    link
    fedilink
    English
    arrow-up
    2
    ·
    edit-2
    7 hours ago

    Atm the meta for local inference is unified memory and “routed” local agents (multiple smaller role-specific agents in a trench coat)

    The former is standout for cost efficiency (e.g., 4x RDMA 48gb Minis for a 192gb cluster @ $43/gb vs a $45k b200 alone)

    The latter is standout for many things, including resource efficiency on smaller machines

    • Nouvellalia@lemmy.world
      link
      fedilink
      English
      arrow-up
      1
      ·
      3 hours ago

      That’ll surely work to automate simple mental labor or tasks, but I don’t think a swarm of minis is going to be able to provide impactful personal and psychosocial analysis of your life.