Possibly linux@lemmy.zip to MeanwhileOnGrad@sh.itjust.worksEnglish · 15 days agoMe poking Deepspeak-r1 with a stick...lemmy.zipimagemessage-square27fedilinkarrow-up1106
arrow-up1106imageMe poking Deepspeak-r1 with a stick...lemmy.zipPossibly linux@lemmy.zip to MeanwhileOnGrad@sh.itjust.worksEnglish · 15 days agomessage-square27fedilink
minus-squareboredtortoise@lemm.eelinkfedilinkEnglisharrow-up1·15 days agoYes. Every model has its overlords altering information. It’d be fine if models would be only instructed to avoid harmful content generation on ethical grounds but it’s quite different from being developed to produce fabrications on purpose.
Yes. Every model has its overlords altering information. It’d be fine if models would be only instructed to avoid harmful content generation on ethical grounds but it’s quite different from being developed to produce fabrications on purpose.