haxor@derp.fooMB to Hacker News@derp.fooEnglish · 3 years agoTransformers struggle with generalizing tasks beyond pre-training dataarxiv.orgexternal-linkmessage-square2fedilinkarrow-up12arrow-down10file-text
arrow-up12arrow-down1external-linkTransformers struggle with generalizing tasks beyond pre-training dataarxiv.orghaxor@derp.fooMB to Hacker News@derp.fooEnglish · 3 years agomessage-square2fedilinkfile-text
minus-squareAggressivelyPassive@feddit.delinkfedilinkEnglisharrow-up2·3 years agoWell, who would have thunk?
minus-squarenautilus@lemmy.dbzer0.comlinkfedilinkEnglisharrow-up1·3 years agoThat and Starscream is being very uncooperative