spaduf@slrpnk.net to Technology@lemmy.worldEnglish · 6 days agoDeepSeek releases new image model familytechcrunch.comexternal-linkmessage-square15fedilinkarrow-up1161arrow-down19cross-posted to: technology@lemmy.world
arrow-up1152arrow-down1external-linkDeepSeek releases new image model familytechcrunch.comspaduf@slrpnk.net to Technology@lemmy.worldEnglish · 6 days agomessage-square15fedilinkcross-posted to: technology@lemmy.world
minus-squarethedarkfly@feddit.nllinkfedilinkEnglisharrow-up7·6 days agoEven though it’s magnitudes lower than comparable models, Deepseek still cost millions to train. Unless someone’s willing to invest this just to retrain it from scratch, you’re left with the alignment of its trainers.
minus-squarePhoenixz@lemmy.calinkfedilinkEnglisharrow-up1·6 hours agoGood point. Is the training set malleable, though? Could you give it some additional rules to basically sidestep this?
Even though it’s magnitudes lower than comparable models, Deepseek still cost millions to train. Unless someone’s willing to invest this just to retrain it from scratch, you’re left with the alignment of its trainers.
Good point.
Is the training set malleable, though? Could you give it some additional rules to basically sidestep this?