Kubuyekezwa nsuku zonke2547 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.
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 izindabaImboni
I-NVIDIA ihlola kuqala uhlelo lokuphatha amandla lokwengeza umthamo we-AI phakathi kwesabelomali esingaguquki sesikhungo sedatha
I-NVIDIA ithi i-DSX MaxLPS suite yayo ingakwazi ukubuyisela amandla e-rack angasetshenzisiwe, ithuthukise ukusebenza kwe-watt ngayinye futhi isekele umthamo we-Rubin GPU ofinyelela ku-40% ngaphakathi kwesabelomali samandla esifanayo. Isofthiwe isesekubukeni kuqala konjiniyela, futhi izibalo zokusebenza zivela ekuhloleni komthwalo wokusebenza omele i-NVIDIA uqobo.
developer.nvidia.comUkuqamba
I-Forbes ibika ukuthi ihhanisi le-AVO le-Nvidia lisize i-Claude i-Opus 5 yafinyelela ku-100% ku-ARC-AGI-3
I-Forbes ibika ukuthi i-Nvidia's AVO's harness harness inyuse izinga lokuqedwa kwe-Claude Opus 5 kusethi yomphakathi ye-ARC-AGI-3 ukusuka cishe ku-30% ukuya ku-100% ngaphandle kokushintsha izisindo zemodeli. Umphumela uthuthukisa umbiko okhona mayelana nesistimu ye-AVO ye-Nvidia, kodwa izimangalo ze-athikili enikeziwe aziqinisekiswanga ngokuzimele...forbes.comUkuphepha
I-NVIDIA ithi ukuphepha komenzeli we-AI kufanele kuhlale ngaphansi kwehhanisi
I-NVIDIA ithi izilawuli zokuphepha ze-AI-ejenti kufanele ziphoqelelwe izikhathi zokusebenza nengqalasizinda, kunokuba kusetshenziswe imodeli eguquguqukayo noma i-harness logic. Ibhulogi yayo yezobuchwepheshe iphakamisa ukwakheka okunezingqimba kanye namaphrofayili amane omthwalo wokusebenza, kodwa ayinikezi ukuqinisekiswa okuzimele kwendlela.developer.nvidia.comUkuqamba
I-Google DeepMind ihlela ucwaningo lwe-AI esiteji socwaningo lwe-AI emhlabeni wegeyimu ophikelelayo we-EVE
I-Google DeepMind ithi inweba ucwaningo lwayo lwemidlalo ngokubambisana ne-Fenris Creations, isitudiyo esilandela umkhathi we-EVE. Uhlelo luzohlola inkumbulo yama-ejenti, ukufunda okuqhubekayo, ukuhlela isikhathi eside nokuziphatha kwama-ejenti amaningi, kuqala ngendawo engaxhunyiwe ku-inthanethi ye-EVE.
deepmind.googleUkuphepha
Uhlelo lokusebenza lomthombo ovulekile luxwayisa mayelana nezibuko ezihlakaniphile eziseduze, kodwa amasiginali akhona angummeleli onzima kuphela
Iphrojekthi ye-Nearby Glasses isebenzisa idatha ye-Bluetooth ukumaka amanye amadivayisi anezibuko ezihlakaniphile ku-Android ne-iOS. Imibhalo yayo ithi lolu hlelo lungakhiqiza izinto ezingamanga, izinto ezigejayo futhi akufanele zisetshenziselwe ukubhekana nabantu.github.comUkuphepha
I-Google i-Threat Intelligence ibika ngocwaningo lobungozi olusizwa yi-AI kanye nohlelo olungayilungele ikhompuyutha oluzimele
I-Google Threat Intelligence Group ithi izitha zisebenzisa i-AI ukusheshisa ucwaningo lokuba sengozini, ukubalekela ukutholwa, ukwenza ngokuzenzakalelayo izenzo zohlelo olungayilungele ikhompuyutha futhi kuqondiswe kuchungechunge lokuhlinzeka ngesofthiwe ezizungeze izinhlelo ze-AI.cloud.google.comUkuphepha
Imibiko yokuphrinta kusengaphambili imiphumela yemodeli yolimi ingavuza izimfihlo ezigcinwe kumongo
Ukuphrinta kwangaphambili kwe-arXiv kubika ukuthi amamodeli olimi angase aveze idatha ebucayi ebekwe kumongo, ngisho nangemva kokwenqaba izicelo eziqondile zokukhipha. Ekuhlolweni okulawulwayo, ababhali bathi ukuhlasela okuguquguqukayo kubuyise izimfihlo ezimfushane kusuka kokuphumayo okuvamile futhi kwakhipha izinombolo ezigcwele Zokuvikeleka Komphakathi kumenzeli wesitayela sokukhiqiza.arxiv.orgUkuqamba
Ibhentshimakhi yebhokisi lokucabanga ithola ama-agent e-AI azabalaza ukuze aqedele ukusebenza kahle kwebhizinisi okuthembekile
Iphepha elisha le-arXiv lethula i-Thinkbox, i-sandbox kanye ne-benchmark yokuhlola ama-agent e-AI emisebenzini yebhizinisi enezinyathelo eziningi. Imodeli evivinywe eqine kakhulu ithole amaphuzu wokudlula angu-65.36% kodwa yaphumelela ngo-25.25%^20 amaphuzu, okugqamisa igebe phakathi kwempumelelo yezikhathi ezithile nokwenza okunokwethenjelwa.arxiv.orgUkuqamba
I-Preprint ibika ubungozi be-asymmetric amamethrikhi wokuminyanisa we-LLM angakwazi ukuzigeja
Iphrinta ye-arXiv ehlola ama-LLM angu-3 kuzo zonke izindlela zokuminyanisa ezingu-11 ibika ukuthi ukunemba okuhlanganisiwe kanye nokudideka kungafihla ukulahlekelwa kolwazi olungalingani, ukuzethemba ngokweqile olwazini olusanda kulahleka, kanye nokunciphisa amashifu ekuchemameni kweqembu elincane.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