Kwenzekeni
I-TechCrunch ibika ukuthi i-Google DeepMind kanye ne-Google Research ikhiphe i-WeatherNext 3, imodeli ye-AI yokubikezela isimo sezulu ezoqala ukuhlinzeka ngolwazi lwesimo sezulu lwe-Google Usesho, i-Maps, ne-Gemini, kuyilapho lwenziwa lutholakale kubasebenzisi bengxenyekazi ye-ZXQAIU yesikhulumi se-ZXQAIU.
I-TechCrunch ibika ukuthi i-WeatherNext 3 ingabikezela okuguquguqukayo okubalulekile, okuhlanganisa izinga lokushisa, isivinini somoya, nomswakamo, ngokulungiswa okungaba amakhilomitha angu-5. Le nkampani iphinde ibike ukuthi ukuhlolwa kwayo kwemvula kuthuthuke ngo-60% uma kuqhathaniswa ne-WeatherNext 2 nokuthi imodeli ingakwazi ukukhiqiza izibikezelo zehora kunezikhawu ezivame kakhulu zamahora ayisithupha. I-Google ayibhalanga ukutholakala komthengi okuthile noma intengo emthonjeni.
Ngokusho kwe-TechCrunch, i-WeatherNext 3 inkulu ngokuphindwe ngo-2.4 kuneyandulelayo futhi yaqeqeshelwa ukubikezela izimo eziteshini ezithile zesimo sezulu, isiza ukuxhumanisa ukuqagela nokubonwa okusekelwe phansi. Imodeli iphinde ingenise idatha yesimo sezulu sesathelayithi eqoqwe njalo ngehora. I-TechCrunch ibika ukuthi i-Google ichaza lokhu njengemodeli yokuqala ye-AI yokusebenzisa ngokuqondile ukubonwa okungaphekiwe kwesibikezelo somhlaba wonke sokulungiswa okuphezulu, nakuba i-WindBorne iwuphikisa lowo mehluko futhi ithi imodeli yayo ye-WeatherMesh 6 isebenzise ukubonwa okungavuthiwe kusukela ngasekupheleni kuka-2025. Umthombo uthi zombili izinhlelo zisathembele kudathasethi eqondile yesimo sezulu egcwele kazwelonke.
Imininingwane yomthombo: techcrunch.com ↗
Kungani kubalulekile
Izibikezelo ezinokulungiswa okuphezulu nezivamile zingenza ulwazi lwesimo sezulu olusekelwe ku-AI lusebenziseke kakhulu ekuhleleni kwendawo, ukusebenza kwamandla avuselelekayo, ezolimo, kanye nezifunda ezingakwazi ukukhokhela ingqalasizinda ejwayele ukusetshenziselwa ukubikezela isimo sezulu sezinombolo. Imiphumela ebikiwe isalokhu iyizimangalo ezivela ku-Google nokuhlola kwebhentshimakhi okuchazwe yi-TechCrunch, hhayi ukuqinisekiswa okuzimele.
Izinguquko ezibikiwe zibhekana nobuthakathaka obuthile obusebenzayo bezinhlelo zesimo sezulu ze-AI: ukulungiswa kwendawo okumaholo, izibikezelo zemvula ezibuthakathaka, izibuyekezo ezingavamile, nokuthembela kumadathasethi kahulumeni asesetshenzwe ngaphambili. Ukuqagela okwenziwe okwasendaweni okwengeziwe nokuvamile kungase kuthuthukise izinqumo ezincike emoyeni, emvuleni, nasembozweni yamafu, okuhlanganisa ukuhlela amandla avuselelekayo nemisebenzi yezolimo.
I-TechCrunch ithi i-WeatherNext 3 yenze kangcono kunezinye izinhlobo zokufunda ezijulile ezivela ku-Google, Microsoft, Nvidia, kanye ne-European Centre for Medium-Range Weather Forecasts, kanye nezibikezelo zendabuko ezivela ku-U.S. National Weather Service kanye ne-ECMWFBench, ku-Operational Weather Leyo miphumela ayiqinisekiswanga ngokuzimela lapha, futhi umthombo awunikezi amaphuzu ayisisekelo, izikhathi zokuhlolwa, ukuhlukaniswa kwezindawo, noma imininingwane edingekayo ukuze kuhlolwe ukuthi inzuzo isebenza kabanzi kangakanani.
I-Interactive Mechanism: Indlela Esebenza Ngayo Ngempela
Hlola ubuchwepheshe obuyisisekelo ngemuva kwalokhu kuthuthukiswa ngokuhlanganyela.
Which component of an AI application is the machine-learning model itself?
Ongakubuka ngokulandelayo
Buka ukukhishwa kwangempela kwemodeli kuyo yonke imikhiqizo ye-Google, imigomo namanani okufinyelela amafu, nokuhlolwa okuzimele kokusebenza emvuleni enzima yendawo kanye nezimo zezulu ezimbi kakhulu.
Umbuzo obalulekile osebenzayo ukufinyelela. I-TechCrunch ibika ukuthi i-WeatherNext 3 izophakela imikhiqizo ye-Google futhi itholakale nge-Google Cloud, kodwa i-athikili ayisunguli ukuthi ukuhlanganiswa ngakunye kuzovela nini, ukuthi iyiphi abasebenzisi abazoyithola, yimaphi ama-API noma amasevisi ahilelekile, noma ukuthi kuzobiza malini ukufinyelela kwamafu.
Ukucutshungulwa okwengeziwe kufanele kugxile ezenzakalweni zemvula enamandla, izivunguvungu, ukusebenza kwesifunda, ukulinganiswa kwesimo sezulu, nokuthi ingabe ukulungiswa okuphezulu kwemodeli kuhumushela ezinqumweni ezingcono kunemiphumela enemininingwane eminingi nje. Umehluko phakathi kokungenisa kokubuka okungaphekiwe kanye nokulinganisa kwedatha okuqondile nakho kusalokhu kuwumkhawulo obalulekile wezobuchwepheshe.