Basically the title.

For me, a dense model around 20b, with an additional engram table of 10-20b, or a sparse MoE around 80b would be amazing.

  • fluxx@lemmy.world
    link
    fedilink
    English
    arrow-up
    6
    ·
    1 month ago

    What are the significant changes? Requiring attribution or something more sinister?