Lugh@futurology.todayM to Futurology@futurology.todayEnglish · 1 month agoIt's makers claim that an AI which can write the code to improve itself, more than doubled its performance In benchmarks like SWE-bench and Polyglot, outperforming hand-designed agents.sakana.aiexternal-linkmessage-square0linkfedilinkarrow-up11arrow-down10
arrow-up11arrow-down1external-linkIt's makers claim that an AI which can write the code to improve itself, more than doubled its performance In benchmarks like SWE-bench and Polyglot, outperforming hand-designed agents.sakana.aiLugh@futurology.todayM to Futurology@futurology.todayEnglish · 1 month agomessage-square0linkfedilink
