Remix.run Logo
c7b 4 hours ago

Interesting. Is there a paper that explains this in a bit more detail, like [0] for abliteration (underlying the Heretic software, afaik)?

[0] https://arxiv.org/abs/2406.11717

wren6991 4 hours ago | parent [-]

That paper you linked has all of the information you need. The linear algebra they do on the weights there to null out one direction reduces to a single (dot-product + broadcast-multiply + add) on the activations.