- I-Amuse 3.1 inika amandla ukudalwa kwezithombe ezikhiqizwe yi-AI ngokuqondile nasendaweni kumakhompyutha aphathekayo anama-AMD Ryzen AI processors ane-XDNA 2 NPU.
- Isebenzisa imodeli ye-Stable Diffusion 3 Medium FP16/BF16, elungiselelwe ukuthuthukisa ikhwalithi nokunciphisa ukusetshenziswa kwememori kufika ku-9 GB kumakhompyutha ano-24 GB we-RAM.
- Uhlelo luletha ukulungiswa kokugcina kwe-4MP (2048x2048) ngenxa yepayipi elinezigaba ezimbili elenza ukukhushulwa okuhlanganisiwe ku-NPU.
- Ukusebenzisana phakathi kwe-AMD kanye ne-Stability AI kushayela amakhono athuthukile e-AI endawo, okuphawula uguquko ekucutshungulweni kwamafu kanye nokwenza ngcono ubumfihlo nokusebenza kahle.
Ukusebenzisana kwe-AMD kanye nokuzinza kwe-AI kuthatha isinyathelo esibalulekile ekuhlanganisweni kobuhlakani bokwenziwa okukhiqizayo kumakhompyutha womuntu. Kuze kube manje, ukuthuthuka okuningi ku-AI ekudalweni kwezithombe kuncike efwini, kodwa inguqulo entsha ye-Amuse 3.1 iguqula leli qiniso ngokuvumela i ukukhiqizwa kwesithombe esingaxhunyiwe ku-inthanethi ngokuphelele kanye nokwehliswa kabusha kumalaptop afakwe ama-AMD Ryzen AI processors kanye ne-XDNA 2 NPU yayo.
Lobu buchwepheshe Ithulwa njengesixazululo esenzelwe ochwepheshe, abadali nabaqambi abafuna ukwenza imisebenzi ye ukukhiqizwa kwezithombe zekhwalithi ephezulu ngaphandle kokuncika kumasevisi angaphandle noma ubhekane nemikhawulo yoxhumano lwe-inthanethi. Iphinde ithuthukise ubumfihlo kanye nokushesha, njengoba yonke inqubo yenzeka kukhompyutha yomsebenzisi.
I-AI ekhiqizayo yendawo ngempela
Nge-Amuse 3.1, amakhompyutha aphathekayo ahlanganisa I-AMD Ryzen AI 300 noma i-AI MAX+ ingafinyelela ekudalweni kokuqukethwe okubukwayo isebenzisa imodeli entsha Ukusabalalisa Okuzinzile 3 Okumaphakathi ngefomethi ye-FP16 noma ye-BF16. Lokhu kulungiselelwa kuvumela i- ukukhiqizwa kwesithombe sendawo sifinyelele isinqumo se Amaphikseli we-2048 x 2048 (4MP) kusukela kusisekelo sokuqala samaphikseli angu-1024 x 1024, konke sibonga ipayipi lezigaba ezimbili eliqala ngokukhiqiza isithombe bese liyasikala, ku-NPU ngokuphelele.
Uhlelo lunciphisa kakhulu ukusetshenziswa kwenkumbulo uma kuqhathaniswa nokusetshenziswa kwangaphambilini. Nakuba Kunconywa ukuthi ube no-24 GB we-RAM, imodeli ngokwayo isebenzisa kuphela cishe u-9 GB ngesikhathi sokukhiqiza, okuvumela lokhu okuhlangenwe nakho kuma-laptops wengxenye ye-premium, ngaphandle kokudinga ukudlula imemori engu-32 GB.
Ngaphezu kwalokho, ukuhlanganiswa kwemodeli ye-FP16/BF16 kunikeza ibhalansi phakathi ukunemba, ikhwalithi yombala kanye nokusebenza, izimisa phakathi kwezibalo ze-FP16 zendabuko kanye nokusebenza okuphezulu kwe-INT8, kuyilapho igwema ukulahlekelwa kwekhwalithi okuvamile nokulinganisa okuningi. Abasebenzisi bangafinyelela imiphumela ebonakalayo enemininingwane, eyenziwe ngokwezifiso, ifanele ngisho nokuphrinta kochwepheshe.
Izinzuzo nezimfuneko zabasebenzisi abanobuhlakani
I-Amuse 3.1 entsha iqondiswe ikakhulukazi kulabo abayidingayo Dala izithombe zesitoko, izinsiza zokuklama izithombe, noma okubonwayo okuqondene nomkhiqizo othile.. Umsebenzisi angaqala isizukulwane ngokulula ukwaziswa kombhalo, ezizwela kakhulu isakhiwo, ukwakheka, nemininingwane ebhaliwe. Izinguquko ezincane ekunembeni kokwaziswa zinomthelela omkhulu kumphumela wokugcina.
Imodeli isekela izinketho ezithuthukisiwe, njengokwaziswa okungekuhle kokukhipha izinto ezingafuneki noma ukulungiselelwa okuthile ukuze kutholwe izithombe ezithembeke kakhudlwana kuzimfuneko zephrojekthi ngayinye. Futhi Isiza ukuphindaphinda ngokushesha, ivumela ukwahluka okuhlukile ukuthi kwenziwe ngokushintsha imbewu noma amapharamitha. bese ukhetha inketho efaneleke kakhulu yesidingo ngasinye.
Ukuze ufinyelele le misebenzi, udinga nje ikhompuyutha ephathekayo ene I-Ryzen AI 300/AI MAX+, XDNA 2 NPU okungenani enama-TOPS angama-50 kanye nobuncane be-RAM engu-24GB. I-Amuse 3.1 iyatholakala njengokulandwa kwe-beta kwa-Tensorstack, futhi kuyatuswa ukuthi kufakwe abashayeli bakamuva be-AMD Adrenalin ukuze kusetshenziswe ngokugcwele i-NPU. Ngaphakathi kohlelo lokusebenza, umsebenzisi kufanele enze imodi ye-HQ isebenze futhi anike amandla inketho ye-'XDNA 2 Stable Diffusion Offload' ukuze yonke inqubo isebenze endaweni.
Imithelela ye-AI kanye nemboni yobumfihlo
Ukuthuthuka kwe-AMD ku-Stability AI kuphawula isikhundla esicacile uma siqhathaniswa nabanye abakhiqizi, abafana ne-NVIDIA ne-Intel, nabo abagxile ekuthuthukiseni izixazululo ze-AI ezikhiqizayo zasendaweni. Indlela ye-AMD igxile kukho khulisa ukukhiqizwa kokuqukethwe, ukukhulisa isivinini, ukusebenza kahle kanye nobumfihlo, kanye nokwethula a izinga elisha embonini: kungenzeka ukuthi yenza imisebenzi eyinkimbinkimbi ye-AI, njengokwenza izithombe zekhwalithi ephezulu, ngaphandle kokuthembela kumaseva akude noma ukuxhumana namafu.
La Ilayisensi yomphakathi ye-AI yokuzinza ilawula ukusetshenziswa kwemodeli, ivumela ukuxhashazwa kwayo mahhala kubantu ngabanye, amaphrojekthi ocwaningo, namabhizinisi amancane anenani elingaphansi kwesigidi esingu-$1 ngonyaka. Ezinhlanganweni ezinkulu, Imodeli yebhizinisi itholakala ngaphansi kwelayisensi ethile.
Lokhu kunyakaza kwenzeka kumongo omakwe ngokuthi izimpikiswano zomthetho mayelana ne-copyright kanye nokusetshenziswa kwedatha ukuqeqesha amamodeli IA, izinselele i-Stability AI ebhekana nazo ngokunikeza ukukhanya okukhulu nokuphathwa kwelayisensi elawulwayo, kuyilapho ihlukanisa ikhathalogi yayo yamathuluzi okukhiqiza ngezinhlelo zokusebenza ezisesithombeni, ividiyo, nomsindo.
I-AMD kanye ne-Stability AI ibonisa nge-Amuse 3.1 ukuthi Ukukhiqizwa kwesithombe kwe-AI okuthuthukisiwe manje kuyasebenza ezindaweni zasendaweni, ukuhlanganisa ukusebenza kahle, ikhwalithi, nokulawula inqubo yokudala, okuvumela ukuzimela okukhulu kobuchwepheshe nokuphepha kwedatha.
Ngingumshisekeli wezobuchwepheshe oguqule izintshisekelo zakhe "ze-geek" zaba umsebenzi. Ngichithe iminyaka engaphezu kwengu-10 yempilo yami ngisebenzisa ubuchwepheshe obusezingeni eliphezulu kanye nokukitaza ngazo zonke izinhlobo zezinhlelo ngenxa yelukuluku lokufuna ukwazi. Manje sengiqeqeshelwe ubuchwepheshe be-computer nemidlalo yama-video. Lokhu kungenxa yokuthi sekuphele iminyaka engaphezu kwengu-5 ngisebenza ngokubhalela amawebhusayithi ahlukahlukene ezobuchwepheshe nemidlalo yevidiyo, ngenza izindatshana ezifuna ukukunikeza imininingwane oyidingayo ngolimi oluqondakala yiwo wonke umuntu.
Uma unemibuzo, ulwazi lwami lusukela kuyo yonke into ehlobene nesistimu yokusebenza ye-Windows kanye ne-Android yomakhalekhukhwini. Futhi ukuzibophezela kwami kuwe, ngihlala ngizimisele ukuchitha amaminithi ambalwa futhi ngikusize uxazulule noma yimiphi imibuzo ongase ube nayo kulo mhlaba we-inthanethi.