• OozingPositron@feddit.cl
    link
    fedilink
    English
    arrow-up
    3
    ·
    9 months ago

    Basically Mistral, check /lmg/ in /g/, if you have a GPU newer than 2 years you can probably run a 32B quantised model.