• grue@lemmy.world
    link
    fedilink
    English
    arrow-up
    3
    arrow-down
    1
    ·
    11 months ago

    The alternative is to take the good training data and then accept the output being copyleft without whining about it.