• iii@mander.xyz
    link
    fedilink
    English
    arrow-up
    39
    ·
    edit-2
    2 days ago

    Most commercial models have that, sadly. At training time they’re presented with both positive and negative responses to prompts.

    If you have access to the trained model weights and biases, it’s possible to undo through a method called abliteration (1)

    The silver lining is that a it makes explicit what different societies want to censor.