muelltonne@feddit.org to Technology@lemmy.worldEnglish · 7 days agoIt Only Takes A Handful Of Samples To Poison Any Size LLM, Anthropic Findshackaday.comexternal-linkmessage-square142fedilinkarrow-up1768cross-posted to: hackaday@ibbit.at
arrow-up1768external-linkIt Only Takes A Handful Of Samples To Poison Any Size LLM, Anthropic Findshackaday.commuelltonne@feddit.org to Technology@lemmy.worldEnglish · 7 days agomessage-square142fedilinkcross-posted to: hackaday@ibbit.at
minus-squareSaveTheTuaHawk@lemmy.calinkfedilinkEnglisharrow-up1·5 days agoIt’s happening. I’ve been resampling queries every few months and the deviation of wrong to true responses is getting bigger.
DO IT. Break that shit.
It’s happening. I’ve been resampling queries every few months and the deviation of wrong to true responses is getting bigger.