Otter@lemmy.ca to Programming@programming.devEnglish · 2 years agoAnthropic can now track the bizarre inner workings of a large language modelwww.technologyreview.comexternal-linkmessage-square7linkfedilinkarrow-up154arrow-down12cross-posted to: localllama@sh.itjust.workstechnology@lemmy.zipai_@lemmy.world
arrow-up152arrow-down1external-linkAnthropic can now track the bizarre inner workings of a large language modelwww.technologyreview.comOtter@lemmy.ca to Programming@programming.devEnglish · 2 years agomessage-square7linkfedilinkcross-posted to: localllama@sh.itjust.workstechnology@lemmy.zipai_@lemmy.world
minus-squarerecursiveInsurgent@lemm.eelinkfedilinkEnglisharrow-up4arrow-down2·2 years agoInteresting how these findings refute the assertion that LLMs are just predicting the next word. Sometimes they plan ahead.
Interesting how these findings refute the assertion that LLMs are just predicting the next word. Sometimes they plan ahead.