Kwenzekeni
Ihhovisi likaMongameli Wezimboni Zokuvikela (SSB) limemezele ukwethulwa kwe-EVREN, inkundla yezobunhloli kazwelonke ehloselwe umkhakha wezokuvikela waseTurkey. Inkundla ihlanganisa amasethi edatha, amamodeli e-AI kanye neqoqo le-GPU elisebenza kahle kakhulu kuphothali eyodwa (evren.ssyz.org.tr) futhi idinga ukuqinisekiswa kwe-e‑Government (e‑Devlet) ukuze ifinyelele. Kunokukhokhisa izinkokhelo, i-EVREN isebenzisa imodeli yekhredithi: abasebenzisi bathola amakhredithi ngokulayisha noma ukulebula idatha nangokwabelana ngamamodeli aqeqeshiwe, okungase kusetshenziselwe ukuqeqeshwa kwamamodeli kanye nencazelo yemodeli yolimi olukhulu. Isevisi ihlanganisa amathuluzi okubona kokuphela ukuya ekupheleni kwekhompuyutha (ukutholwa kwento, ukuhlukaniswa, ukuhlukaniswa) kanye nesendlalelo se- sama-LLM angu-11 anesisindo esivulekile. Amakholi e-API enziwe ngaphambi komhla ka-1 Nov 2026 akhululwa emalini ebanjwayo. Esikhathini esifushane, inkundla ibike abasebenzisi abasebenzayo abangu-7,500. Izinhlelo zesikhathi esizayo zihlanganisa ukungeza ama-LLM athuthukiswe ekhaya, amamodeli wezithombe nomsindo, nokwandisa umthamo we-GPU ngendlela yokuphatha esabalalisiwe.
NgeSonto, i-SSB ikhiphe isitatimende esiqinisekisa ukwethulwa kokusebenza kwe-EVREN, inkundla ye-AI kazwelonke eyenzelwe amacala okusetshenziswa ahlobene nezokuvikela. Inkundla ihlanganisa idatha, amamodeli e-AI kanye nephuli ye-GPU esebenza kakhulu ngaphansi kwephothali yewebhu eyodwa, efinyeleleka ngesistimu yokuqinisekisa ye-e‑Government yaseTurkey.
Imodeli yokufinyelela ye-EVREN isekelwe kumnikelo: abasebenzisi bathola amakhredithi ngokunikela ngamasethi edatha, ukulebula idatha, noma ukwabelana ngamamodeli aqeqeshiwe. Lawa makhredithi angasetshenziswa emisebenzini ejulile ye-GPU njengokuqeqeshwa kwemodeli kanye nencazelo ngamamodeli olimi amakhulu angu-11 anesisindo esivulekile. Inkundla futhi inikezela ngamapayipi okubona ngekhompyutha ukuze kutholwe into, ukuhlukaniswa nokuhlukaniswa.
Ukuze kugqugquzelwe ukutholwa kusenesikhathi, i-SSB imemezele ukuthi amakholi e-API enziwe ngaphambi komhla ka-1 Nov 2026 ngeke adonswe kumabhalansi esikweletu sabasebenzisi. Ngaphakathi kwewindi elifushane lokuqalisa, inkundla ibike abasebenzisi abasebenzayo abangu-7,500, okubonisa ukutholwa okusheshayo phakathi kwamafemu okuvikela, izinkampani zobuchwepheshe, izifundiswa nabafundi.
Izinhlelo zentuthuko zesikhathi esizayo zihlanganisa ukuhlanganisa ama-LLM adalwe ekhaya, amamodeli wezithombe nomsindo, nokwandisa ingqalasizinda ye-GPU ngokusebenzisa indlela yokuphatha esabalalisiwe.
Imininingwane yomthombo: dailysabah.com ↗
Kungani kubalulekile
I-EVREN imele umnyakazo we-Turkey wokwakha ingqalasizinda ye-AI ezimele yomkhakha wayo wezokuvikela, wehlise ukuthembela kumasevisi wamafu angaphandle angadalula idatha ebucayi. Ngokugcina ukucutshungulwa kwedatha kuzingxenyekazi zekhompuyutha zasekhaya ze-GPU, inkundla ibhekana nokukhathazeka kwezokuphepha kuzwelonke kuyilapho ikhuthaza i-ecosystem ehlanganyelwayo lapho abanikeli bekhuthazwa ukwabelana ngezinsiza. Imodeli esekelwe kukhredithi yehlisa izithiyo zokungena zamafemu amancane, izifundiswa nabafundi, okungenzeka kusheshise ukuthuthukiswa kwethalente le-AI kanye nokuqanjwa okusha ezweni. Ngaphezu kwalokho, ama-LLM esisindo esivulekile wenkundla namandla okubona ngekhompyutha angasheshisa ukuthuthukiswa kwe-prototype yezinhlelo zokusebenza zokuzivikela, kusukela kumasistimu azimele kuya ekuhlaziyweni kobuhlakani. Nokho, ukuntuleka kwamanani entengo adalulwe esidlangalaleni, amabhentshimakhi wokusebenza kwemodeli, nokuqinisekiswa kwangaphandle kusho ukuthi umthelela ongokoqobo wenkundla usazobonakala.
Ingqalasizinda ye-AI Ebusayo inciphisa ubungozi bedatha yezokuvikela ebucayi ecutshungulwa ezinsizeni zamafu zangaphandle, ziqondaniswe nezinto ezibalulekile zokuphepha kuzwelonke.
Isistimu yokunikela esekelwe kukhredithi yehlisa izithiyo zezimali, okungenzeka kube nokufinyelela kwentando yeningi ezinsizeni ze-AI eziphezulu zamabhizinisi amancane namaqembu ocwaningo.
Ngokuhlanganisa idatha namamodeli, i-EVREN ingasheshisa ukuthuthukiswa kobuchwepheshe bokuvikela obushayelwa yi-AI, obufana nezimoto ezizimele, ukuhlaziya okubhekwayo namathuluzi okusekela izinqumo.
Ukuthembela kwenkundla kuma-LLM esisindo esivulekile nokuthuthukiswa kwemodeli yasekhaya kungase kukhuthaze ipayipi lethalente le-AI elikhule ekhaya, kodwa ukungabikho kokuhlolwa kokusebenza okuzimele kushiya ukusebenza kwalawa mamodeli kungaqinisekile.
I-Interactive Mechanism: Indlela Esebenza Ngayo Ngempela
Hlola ubuchwepheshe obuyisisekelo ngemuva kwalokhu kuthuthukiswa ngokuhlanganyela.
crm_get_transaction(id='4092').Which component of an AI application is the machine-learning model itself?
Ongakubuka ngokulandelayo
Izinkomba ezibalulekile okufanele ziqashwe zihlanganisa: (1) amazinga okutholwa ngaphezu kwabasebenzisi bokuqala abangu-7,500, ikakhulukazi phakathi kosonkontileka bezokuvikela nezikhungo zocwaningo; (2) ukukhishwa kwama-LLM akhiwe kuleli nokuthi ayahlangabezana yini namazinga okusebenza afana namamodeli angaphandle; (3) noma yiziphi izinguquko zenqubomgomo kusistimu yekhredithi noma izandiso zesikhathi samahhala se-API; kanye (4) nezilawuli zokuthekelisa ezingaba khona noma ukusabela kwamanye amazwe kuplathifomu ye-AI yokuzivikela.
Ukukhula kwabasebenzisi ngale kwe-7,500 yokuqala, ikakhulukazi phakathi kosonkontileka abasunguliwe, kuzokhombisa ukusebenza kahle kwezentengiselwano kwenkundla.
Ukusebenza nokwamukelwa kwama-LLM akhiwe ekhaya kuzoveza ukuthi iTurkey ingafinyelela ukulingana neminikelo ye-AI yamazwe ngamazwe.
Noma yiziphi izibuyekezo zesistimu yekhredithi noma izandiso zenkathi yamahhala ye-API zingathinta ukusimama kwesikhathi eside kanye nezisusa zabasebenzisi.
Ukusabela kwamazwe ngamazwe, okuhlanganisa izilawuli ezingase zibe khona zokuthumela ngaphandle noma ukukhathazeka kwenxusa, kungase kulolonge ububanzi besikhathi esizayo beplathifomu nokusebenzisana.