[personal][crazy][ot] karl-story/spamlog: heartpulse

k gmkarl at gmail.com
Wed Jan 26 02:15:17 PST 2022


As a brainmind, when doing this, part of the goal is understanding how
it works for reuse.  That's basic to 'algorithm concept optimization'.
So just throwing a transformer model at the problem doesn't reach the
solution unless end-users can fully comprehend and verify the workings
of the model themselves, reapply those concepts, etc, which is a
current research problem. More thoughts rising might give more ideas.
Transformer models could make sense if _we_ understand them better. A
quick explanation for why we aren't using them better is that "we are
mind controlled not to make AGI", this helps a lot of concept areas
move forward with explaining things. It's likely only a quick
explanation and not the real explanation.


More information about the cypherpunks mailing list