sanitation@lemmy.todayBanned from community to Technology@lemmy.worldEnglish · 3 months agoAnthropic just rolled out a tool that'll decimate some people's dreams of writing AI novels undetectedtech.yahoo.comexternal-linkmessage-square32linkfedilinkarrow-up1102arrow-down113cross-posted to: technology@lemmit.online
arrow-up189arrow-down1external-linkAnthropic just rolled out a tool that'll decimate some people's dreams of writing AI novels undetectedtech.yahoo.comsanitation@lemmy.todayBanned from community to Technology@lemmy.worldEnglish · 3 months agomessage-square32linkfedilinkcross-posted to: technology@lemmit.online
minus-squareMangoCats@feddit.itBannedlinkfedilinkEnglisharrow-up7arrow-down1·edit-22 hours agoRemoved by mod
minus-squareKairos@lemmy.todaylinkfedilinkEnglisharrow-up1·3 months agoThat’s not how the watermarking works. It depends on a pattern deviations from statistics throughout a string of tokens.
minus-squareKairos@lemmy.todaylinkfedilinkEnglisharrow-up1·3 months agoI don’t know much about it but it sounds like it would only work if you have the weights and maybe the running conditions of the model. Otherwise the statistical wright wouldn’t be known. It also sounds really easy to break by changing random words.
Removed by mod
That’s not how the watermarking works. It depends on a pattern deviations from statistics throughout a string of tokens.
Removed by mod
I don’t know much about it but it sounds like it would only work if you have the weights and maybe the running conditions of the model. Otherwise the statistical wright wouldn’t be known.
It also sounds really easy to break by changing random words.
Removed by mod
Thanks :)