Qwen Intelligence Uyasebenza

I-Qwen Intelligence iyisixazululo esiphelele sezingcingo ze-AI esethulwe ithimba le-Alibaba Tongyi Qianwen, esinikeza abakhiqizi bezingcingo amamodeli, inkundla nezixazululo ezisekelwe ezimweni zokusetshenziswa. I-Qwen Intelligence iqukethe ama-Agent amathathu amakhulu: i-Mobile Planner iphethe ukuhlela imisebenzi nokwenza izinqumo, i-Mobile-Use yenza imisebenzi ocingweni, kanti i-Mobile Creative igxile ekudaleni izithombe; ukusebenza kwawo konke kufinyelela ezingeni elihamba phambili embonini. Lesi sixazululo sesisetshenziswe ngokubambisana ne-HONOR, safinyelela ukunemba okuphelele kwemisebenzi okungu-91.8%, okwenza ucingo lusuke ekuphenduleni kuphela luye ekunikezeni isevisi ngokuqhubekayo.

Mobile Planner Agent (Ukuhlela okuhlakaniphile): Njengobuchopho bocingo, iqonda izidingo eziyinkimbinkimbi, yephule imisebenzi ibe izingxenye, ihlele amathuluzi futhi ilungise uhlelo ngokuguquguqukayo, isekele ukulethwa kwemisebenzi emide.
Mobile-Use Agent (Ukusebenza ocingweni): Isebenzisa indlela ethi "API kuqala + GUI njengesipele" ukwenza imisebenzi kuma-app ahlukene, njengokuthola ulwazi nokugcwalisa amafomu, inezendlalelo ezintathu zokulawula ukuphepha nobumfihlo.
Mobile Creative Agent (Ukudala izithombe): Iphula ngokuzenzakalelayo izidingo ezichazwe ngolimi olujwayelekile zibe ukuhamba komsebenzi, ikhiqize isithombe esisebenzisekayo ngomyalo owodwa, kanti ukukhiqizwa kwesithombe sokuqala kuthatha imizuzwana engu-3 kuphela.
Uhlelo lwenkumbulo olunezingqimba ezimbili: Inkumbulo yokusebenza iyashintsha ngokuthuthuka komsebenzi, kanti inkumbulo yesikhathi eside igcina iphrofayela yomsebenzisi, ukuze uhlelo luqonde umsebenzisi kangcono ngokusetshenziswa ngakunye.
Amakhono enkundla aphelele: Luyimodular futhi lungenziwa ngendlela oyifisayo, lusekela ukuthunyelwa kwamamodeli ngokuzimele, ukuphathwa okuhlangene kwama-Skill, ukunakekelwa kwendawo eyodwa nokuhlolwa okuzenzakalelayo.
Uhlelo oluvulekile lokuhlola: Kuvulwe amasethi amane okuhlola, ahlanganisa ukuhlola okuphelele kokuhlela, ukusebenzisa, idivayisi yangempela nokuphepha.

Vakashela iwebhusayithi esemthethweni: Vakashela iwebhusayithi ye-Qwen Intelligence (qwenintelligence.com) ukuze ufunde imininingwane yesixazululo.
Ngena enkundleni ye-Qianwen AI: Vakashela indawo ye-Qwen Intelligence enkundleni ye-Qianwen AI ku-https://www.qianwenai.com/agents/qwenintelligence ukuze uyizame ku-inthanethi.
Sebenzisa imiyalo yolimi olujwayelekile: Veza isidingo ngokuqondile, njengokuthi "ngihlelele uhambo lwebhizinisi oluya e-Shanghai ngoMsombuluko"; i-Planner Agent izohlela ngokuzenzakalelayo futhi yephule umsebenzi ube izingxenye.
Nikeza imvume futhi uqinisekise imisebenzi: Uma kudingeka kuthathwe ulwazi noma kuthunyelwe ifomu, i-Mobile-Use Agent izocela imvume; izinqumo ezibalulekile zenziwa kuphela ngemva kokuqinisekiswa ngumsebenzisi.
Dala umsebenzi ngobuningi bomusho owodwa: Esimweni sokudala, chaza isidingo ngolimi olujwayelekile; i-Creative Agent izokwephula ngokuzenzakalelayo ukuhamba komsebenzi futhi ikhiqize isithombe sokugcina.

Isixazululo esiphelele: Sigxile ekunikezeni abakhiqizi bezingcingo isisekelo esisodwa se-AI esihlanganisa imodeli, inkundla nezixazululo zezimo zokusetshenziswa.
Ukuthuthukiswa okujulile kwezimo zokusetshenziswa: Sisekelwe kumodeli enkulu ye-Qianwen futhi senzelwe ngokukhethekile izimo zezingcingo, sinamandla amakhulu nenani elingcono.
Ukusebenza okuhamba phambili embonini: Ama-Agent amathathu amakhulu afinyelela eqenjini eliphambili noma endaweni yokuqala ekuhlolweni okujwayelekile kokuhlela, ukusebenza nokudala.
Inani eliphakeme kakhulu: Izindleko zokubiza i-Planner Agent emisebenzini eyinkulungwane ngayinye zingamadola aseMelika angu-2.41 kuphela, okuyingxenye encane yezindleko zamamodeli ahamba phambili.
Ukusebenza kahle: Izinga lokuphumelela kwemisebenzi ukusuka ekuqaleni kuye ekugcineni lingu-90%, isikhathi sokuqeda umsebenzi sehliswe ngo-28.8%, kanti izindleko zehliswe ngo-20%.
Iphephile futhi ithembekile: Izendlalelo ezintathu zokulawula ukuphepha zenza amaphuzu okuphepha abe phezulu ngamaphesenti angu-23 kunezixazululo ezihamba phambili, futhi ukuhlolwa kokuphepha kwakhiwa ngokubambisana nethimba le-Fudan Baize.
Iqonda wena kangcono ngokusetshenziswa: Uhlelo lwenkumbulo olunezingqimba ezimbili lugcina iphrofayela yomsebenzisi futhi luqhubeke nokwenza ukubambisana kube okomuntu siqu.

Ukuhlela uhambo lwebhizinisi: Ngomyalo owodwa othi "ngihlelele uhambo lwebhizinisi oluya e-Shanghai ngoMsombuluko", iqhathanisa ngokuzenzakalelayo amanani amathikithi endiza nawesitimela esinesivinini esikhulu, ibhukhe ihhotela owake wahlala kulo futhi isethe izikhumbuzi zohlelo.
Ukugcwalisa ulwazi kuma-app ahlukene: Iqonda izinkomba ezihlakazekile ukuze ithole izithombe ezifanele, ithathe ngokuzenzakalelayo ulwazi igcwalise amafomu, njengamafomu okubuyiselwa kwezindleko noma okubhalisa, bese icela umsebenzisi aqinisekise ngaphambi kokuthumela.
Ukuphathisa imisebenzi emide: Ihlela imisebenzi eyinkimbinkimbi enezinyathelo ezingaphezu kuka-100, njengokuhlela umcimbi, iqoqe ulwazi olusakazekile futhi iqhubekisele phambili ukulethwa kancane kancane.
Ukudala izithombe ngomusho owodwa: Chaza ngolimi olujwayelekile ukuthi "yenza idayari yakudala ehlelwe ngokucacile"; uhlelo luzokwephula ngokuzenzakalelayo ukuhamba komsebenzi futhi lukhiqize isithombe sokugcina ngemizuzwana engu-3, ngaphandle kwesidingo semiyalo yobungcweti.
Ukusebenza esipheqululini nakumadivayisi ahlukene: Amakhono adlulela ezimweni zesiphequluli nasekubambisaneni kwamadivayisi amaningi, kwakheke isevisi ephelele ukusuka ekuqaleni kuye ekugcineni; izinga lokuqedwa komjikelezo wesixazululo se-HONOR selifinyelele ku-90%.