3 ms·
A more concise version of my question is; "How do we train large ML/AI systems to think generating the unholy is bad without hurting intelligence, given we kno
by posting_mess 2y ago
A more concise version of my question is;
"How do we train large ML/AI systems to think generating the unholy is bad without hurting intelligence, given we know that applying some universal law (I.E RLHF) hurts the model".
Trying to promote the exact opposite of "let them eat the unholy".