Kubuyekezwa nsuku zonke2405 izindaba eziqinisekisiwe
Izindaba ze-AI. Ngaphandle komsindo.
Ukufakwa kwe-AI ehlolwe ngomthombo yokwethulwa komkhiqizo, ukushintshwa kwenqubomgomo, ucwaningo lwezokuphepha, nomnyakazo womkhakha, okuchazwe ngesiNgisi esilula ithimba lezemfundo elingenzi nzuzo.
Ingobo yomlando yezindaba
Ingobo yomlando yezindaba ze-AI — Ikhasi 196
Izindaba ze-AI ezihlolwe ngumthombo, ezintsha kakhulu kuqala, zabantu abadinga ukuqonda i-AI ngaphandle kokujaha i-hype.
Ukuthola umthombo oqinisekisiwe
Yonke indaba ixhuma ebufakazini obunamandla obutholakalayo: imithombo yoqobo uma itholakala, ngenye indlela ukubika okuchasiselwe ngokucacile.
IsiNgisi esilula
Kwenzekeni, kungani kubalulekile, nokuthi yini okufanele uyibuke - ngaphandle kwejagoni.
Asikho isigcwalisi
Uma isignali incane, asishicileli lutho kunokupheqa okuphakelayo.
Izindaba ezengeziwe
9 izindabaUkuqamba
Imodeli Ethuthukisiwe Yengqondo Engaqondakali Yokuxilongwa Kwephutha Kwesiguquli Samandla Ngokusebenzisa Ukuthuthukiswa Kwendlela Yegesi Engukhiye ye-IEEE
Lolu cwaningo luveza imodeli ethuthukisiwe ehlanganisa i-Fuzzy Logic kanye ne-IEEE Key Gas Method (FL-KGM) eyethula imisebenzi yobulungu ecocisiwe, amasethi emithetho athuthukisiwe athuthukisiwe, kanye nokwehlukaniswa kwenoveli kwe-CO ne-CO2 ukuze kuqedwe ukungahambisani kokuxilonga.arxiv.orgUkuqamba
Ukuphrinta kusengaphambili okusha kubika izinzuzo ezivela kumamodeli olimi ahlanganisiwe ekushayeni kwamathuluzi ezinyathelo eziningi
Ucwaningo lwe-arXiv luhlola amamodeli olimi ashintshiwe navamile kumabhentshimakhi amathathu okubiza amathuluzi, kubika imiphumela eqinile ekugelezeni komsebenzi okudinga amakholi amaningi e-API ancike kakhulu kanye nendlela esebenza kahle yokuguquguquka yekhompyutha.arxiv.orgUkuqamba
I-Adversarial Review ihlola ukungavumelani okuhleliwe ukuze kubuyekezwe ikhodi ye-ejenti
Iphepha elisha le-arXiv liphakamisa iphrothokholi yokubuyekeza ikhodi yama-ejenti amathathu lapho umbuyekezi ehlola khona ikhodi ye-ejenti kanye nokuhlola okugxekayo okubukezwayo ngaphambi kokuhlelwa. Ababhali babika imiphumela yebhentshimakhi ethuthukisiwe ngaphezu kwesisekelo esihloliwe, kuyilapho bekhomba ukuvumelana okungamanga njengemodi yokwehluleka.arxiv.orgUkuqamba
Ucwaningo lwe-ArXiv lubika izinzuzo ezinkulu zenkulumo-nzondo yesi-Urdu yesi-Roman kusuka ekuzivumelaniseni ne-LoRA
Ukuphrinta ngaphambili kwe-arXiv kuqhathanisa ukushutha kahle okuyiziro kanye nokuhlelwa kahle kwepharamitha ukuze kutholwe inkulumo enenzondo ngesi-Roman Urdu. Ababhali babika ukuthi ukujwayela kwe-LoRA kukhuphule ukusebenza kwe-F1 kusuka ku-0.56 kuya ngaphezulu kuka-0.93 kukhorasi equkethe amazwana angaphezu kuka-72,000 anezichasiselo.arxiv.orgUkuphepha
Ukuhlolwa kokuqala kwe-FraudBench kuthola ama-agent asebhange esengozini yokukhwabanisa okuguquguqukayo
Iphepha le-arXiv lethula i-FraudBench, ibhentshimakhi yokuhlola ukuthi ama-ejenti asebhange asebenzisa amathuluzi angakwazi yini ukubona ukukhwabanisa okwenzeka kuzo zonke izingxoxo. Ekuhlolweni kokuqala kwesilingo esisodwa, ama-ejenti amane athole u-49% kuya ku-65% ekuvikelekeni kokuhlasela.arxiv.orgUkuqamba
Abacwaningi be-Apple babika umthetho wokukala amamodeli okuqeqesha anedatha eyindlala
Ucwaningo lokuqeqeshwa kwemodeli yolimi engaphezu kuka-2,000 luthi idatha eqondiwe entulekayo ingaphinda izikhathi ezingu-15-20 ngezingxube, ngezinga elingcono kakhulu elihluka ngesilinganiso nokubala.machinelearning.apple.comUkuqamba
I-Apple Abacwaningi Baphakamisa Ukushintshwa Kwe-Lexical Ukuze Kuthuthukiswe Ukuqeqeshwa Kwemodeli Yezilimi Eziningi
Abacwaningi be-Apple bachaza i-LINK, ukungenelela kwangaphambi kokuqeqeshwa okungena esikhundleni samagama esiNgisi akhethiwe ngokuhumusha kwezinga legama kusuka olimini okuqondiwe kulo. Iphepha libika intuthuko kuzo zonke izilimi eziyisishiyagalombili namamodeli amamodeli amahlanu, okuhlanganisa ukufika kusivinini esiphindwe kabili ekufinyeleleni ukusebenza okulinganayo komfula.machinelearning.apple.comInqubomgomo
Iphepha lesikhundla lidinga isitifiketi ngaphambi kokuba ama-agent e-AI enze izinqumo zemakethe
Iphepha lesikhundla libika ukungqubuzana okungasho lutho kwabenzeli be-DeepSeek-R1 ekulingiseni imakethe yamanani e-Bertrand, ngisho nangemva kweziyalezo zabantu ngokumelene nokuhlangana. Iphikisa ukuthi ukuqinisekiswa kokuziphatha okubhekiwe kufanele kwandulele ukuthunyelwa kwama-ejenti acabangayo ezimakethe zomnotho; ubufakazi nezivikelo kusalokhu kuyisiqalo.arxiv.orgUkuqamba
Ukubuyekeza okuhlelekile kuveza ukusetshenziswa okukhulayo kwamamodeli olimi olukhulu empilweni yengqondo
Ukubuyekeza okuhlelekile kuhlola ukuthi amamodeli ezilimi amakhulu acwaningwa kanjani ukuze kuhlaziywe impilo yengqondo, ukuhlolwa kwengozi, ukwesekwa kokwelashwa nokuqapha kwezindlela eziningi, kuyilapho kugcizelela izinselele ezingakaxazululwa zezimiso zokuziphatha nokulawula.arxiv.org
Ukwaziswa okukodwa okuwusizo isonto ngalinye
Qhubeka ne-AI ngaphandle kokuhlala kokuphakelayo.
Thola izindaba ze-AI eziqinisekisiwe zeviki, idatha yoqobo, amathuluzi awusizo, ukukhetha okufundayo, nemisebenzi emisha ye-AI.
Finyelela abantu abafunda i-AI
Ukuqasha uchwepheshe we-AI noma ukwethula umkhiqizo owusizo we-AI? Yibeke phambi kwabantu abeze lapha ukuzofunda nokwenza.
Thumela umsebenzi we-AIThumela ithuluzi le-AI