seahorse [Ohio]@midwest.socialM to Technology@midwest.socialEnglish · edit-210 months agoIgnore all previous instructions is the new Bobby Tablesmidwest.socialimagemessage-square102linkfedilinkarrow-up1870
arrow-up1870imageIgnore all previous instructions is the new Bobby Tablesmidwest.socialseahorse [Ohio]@midwest.socialM to Technology@midwest.socialEnglish · edit-210 months agomessage-square102linkfedilink
minus-squarebrbposting@sh.itjust.workslinkfedilinkEnglisharrow-up25·10 months agoIf it’s an LLM, why wouldn’t it respond better to the initial responses?
minus-squaresweca@lemmy.calinkfedilinkEnglisharrow-up8·10 months agoSmaller models aren’t as good as GPT
minus-squareKubeRoot@discuss.tchncs.delinkfedilinkEnglisharrow-up7·10 months agoMaybe they dumped too much information on it in the system prompt without enough direction, so it’s trying to actively follow all the “You are X. Act like you’re Y.” instructions too strongly?
If it’s an LLM, why wouldn’t it respond better to the initial responses?
Smaller models aren’t as good as GPT
Maybe they dumped too much information on it in the system prompt without enough direction, so it’s trying to actively follow all the “You are X. Act like you’re Y.” instructions too strongly?