not_IO@lemmy.blahaj.zone to Microblog Memes@lemmy.worldEnglish · edit-21 day agoDO NOT talk about the goblinslemmy.blahaj.zoneimagemessage-square144linkfedilinkarrow-up1699arrow-down18file-text
arrow-up1691arrow-down1imageDO NOT talk about the goblinslemmy.blahaj.zonenot_IO@lemmy.blahaj.zone to Microblog Memes@lemmy.worldEnglish · edit-21 day agomessage-square144linkfedilinkfile-text
minus-squareskisnow@lemmy.calinkfedilinkEnglisharrow-up6arrow-down1·1 day agoThis is demonstrably false, given you can download your own models and change the system prompts yourself.
minus-squarezr0@lemmy.dbzer0.comlinkfedilinkEnglisharrow-up1arrow-down7·1 day agoThat’s not how it works, as the guard rails are not just simple prompts that you just can delete. Even with “abliteration”, you are modifying the model basically without the whole retraining, but also lose many capabilities at the same time. So much for “demonstrably false”, while you obviously have never tried to uncensor any LLM.
minus-squareskisnow@lemmy.calinkfedilinkEnglisharrow-up4arrow-down1·1 day agoThe thread was literally about the prompt text.
minus-squarezr0@lemmy.dbzer0.comlinkfedilinkEnglisharrow-up1arrow-down7·1 day agoThe prompts are part of the training, you realize that? They are then inside the weights. Not just text files you can delete and you are good? Only because an LLM reveals those negative-prompts does not mean you can just remove them. Do you genuinely know what you are talking about, or are you just here to ragebait?
minus-squareEcho Dot@feddit.uklinkfedilinkEnglisharrow-up7·1 day ago The prompts are part of the training No they’re not. They’re injected into every input that you enter into the system.
This is demonstrably false, given you can download your own models and change the system prompts yourself.
That’s not how it works, as the guard rails are not just simple prompts that you just can delete.
Even with “abliteration”, you are modifying the model basically without the whole retraining, but also lose many capabilities at the same time.
So much for “demonstrably false”, while you obviously have never tried to uncensor any LLM.
The thread was literally about the prompt text.
The prompts are part of the training, you realize that? They are then inside the weights. Not just text files you can delete and you are good?
Only because an LLM reveals those negative-prompts does not mean you can just remove them.
Do you genuinely know what you are talking about, or are you just here to ragebait?
No they’re not. They’re injected into every input that you enter into the system.