3 ms·
Wasn't most of the censoring client-side anyway? Anyway, gow can you make a model truely "unbiased" and have it spit out a quality response?
by ramon156 2y ago
Wasn't most of the censoring client-side anyway? Anyway, gow can you make a model truely "unbiased" and have it spit out a quality response?
- jackdeansmith 2y agoThere was certainly some client side censorship, but also the model weights clearly had some of this baked in. You can try yourself by asking the same queries to r1 hosted by a US host like together.
- NLPaep 2y agoThe distills have this, too. When writing a story with Llama, sometimes including violence will get you a canned refusal paragraph claiming how China is a peaceful place and society.