I-Apple ihlangabezane nokungasebenzi kahle kwengqalasizinda yayo ye-Private Cloud Compute, eyenzelwe ukucubungula imibuzo ye-AI. Amaseva aklanyelwe ngqo le njongo asetshenziswa ngesilinganiso esingu-10% kuphela samandla awo, futhi eminye imishini iqoqa uthuli ezindaweni zokugcina impahla.
Ngokusho 9 ku5Mac, okubhekisela encwadini UlwaziI-Apple iye ku-Google ngesicelo sokusingatha amamodeli ayo amasha e-Siri kumaseva ayo, njengoba ukuhlukana kwezinhlelo zayo zangaphakathi, namaqembu ahlukene e-Apple asebenza ngamaqoqo obuchwepheshe ahlukene, kuvimbela ukusatshalaliswa kabusha kwezinsiza okuguquguqukayo. Ngesikhathi esifanayo, umnyango wezezimali awunelisekile ngezindleko ezikhulayo zengqalasizinda ephindaphindwayo futhi okwamanje awufuni ukutshala izigidigidi ekuyakheni kabusha.
Ngobuchwepheshe, le pulatifomu nayo isalele emuva: amaprosesa e-M2 Ultra aguquliwe awanikezi ukusebenza okudingekayo ukuze kusetshenziswe amamodeli ezilimi ezinkulu (i-LLM), futhi izibuyekezo zesofthiwe zidinga umzamo nesikhathi esikhulu. Ngenxa yalokhu, intshisekelo encane yabasebenzisi ezicini ze-Apple Intelligence ikhulise ukungabaza mayelana nokusetshenziswa okuqhubekayo kwezikhungo zedatha zangaphakathi.
I-Apple iye yacela i-Google ukuthi ifake amamodeli ayo amasha Siri ezikhungweni zayo zedatha, njengoba ingqalasizinda yayo ye-Private Cloud Compute ingakwazi ukusingatha imisebenzi esindayo, kuyilapho i-Google, ngokuphambene nalokho, isivele inolwazi olukhulu ekusetshenzisweni okukhulu kwamaseva e-LLM ngenxa yephrojekthi Gemini.
Source:
Source: 3dnews.ru
