Muse Masked Generative Hoto
Muse samfurin rubutu-zuwa-hoto ne daga Google wanda ke samar da hotuna ta hanyar cike alamun hoto mai rufe fuska gaba daya, yana mai da shi sauri fiye da yada mataki-mataki.
Dubawa
It matters because it showed you can get high-quality, well-aligned images without the slow iterative denoising that most generators rely on.
Zurfafa nutsewa
Muse yana aiki a cikin keɓaɓɓen sararin alama na hoto. VQGAN wanda aka riga aka horar yana juya hoto zuwa grid na alamomin lamba, kamar ƙamus na tubalan ginin gani. A lokacin horo, babban ɓangaren waɗannan alamomin ana rufe su, kuma Mai Canjawa yana koyon tsinkayar su baya, sharadi akan saka rubutu daga daskararre babban samfurin harshe (T5-XXL). A lokacin tsara Muse yana farawa daga grid mai rufe fuska kuma yana yanke hukunci a layi daya, yana tsinkayar alamu da yawa a kowane mataki kuma yana sake rufe mafi ƙarancin ƙarfin gwiwa. Zane-zane mai matakai biyu na farko yana samar da grid mai ƙarancin ƙuduri, sannan samfurin babban ƙuduri ya cika grid mafi girma. Saboda da yawa daga alamu suna warwarewa lokaci guda, ƙirar siga na 900M da 3B suna samar da hoton pixel 256 ko 512 a cikin ɗimbin wucewar gaba.
Fahimtar Fasaha
Babban dabarar shine daidaitawa tare da sake gyara tushen dogaro, wanda galibi ake kira Samfur-style MaskGIT. Maimakon tsinkayar alamar alama ɗaya a lokaci ɗaya (autoregressive) ko ƙin ɗaruruwan lokuta (wasu yaduwa), Muse ya annabta duk alamun da aka rufe, yana kiyaye mafi ƙarfin gwiwa, kuma ya sake rufe sauran don zagaye na gaba. Yin amfani da daskararre T5-XXL mai rikodin rubutu yana ba da ƙarfin fahimtar harshe kyauta, kuma aiki akan sahihan alamu yana ba da damar ƙirar dalili game da hotuna kamar kalmomi.
Dabarun Tasiri
Gudu da sikelin
Kayayyakin AI na iya sarrafa aiki da bincike, ganowa, da ayyuka masu alama a sikelin.
Gina zaɓuɓɓuka
Ƙungiyoyin ƙirƙira za su iya samar da ra'ayoyi cikin sauri tare da ƙarancin bita da hannu.
Ƙungiya da aikin aiki
Ayyuka na iya amfani da siginar hoto da bidiyo waɗanda a baya suke da wahalar aiwatarwa.
Makomar Muse Masked Generative Hoto
Matsakaicin yanke hukunci mai ma'ana zuwa ga janareta waɗanda ke da inganci duka da sauri, waɗanda ke da mahimmanci don daidaita ma'amala da amfani da na'ura. Yi tsammanin ra'ayin tsinkayar alama don haɗawa tare da watsawa da hanyoyin bidiyo mai jujjuyawar kai, da ikon yin fenti, fenti, da gyara mara rufe fuska. Kamar yadda keɓaɓɓen tokenizers ke haɓaka, hoton abin rufe fuska na iya ƙara tsafta zuwa bidiyo da 3D, inda daidaitawa na iya rage farashin samar da firam ko ra'ayoyi da yawa.
Aiwatar da Gaskiyar Duniya
Fahimtar fasaha cikin sauri da allon yanayi inda mai zane yana buƙatar bambance-bambancen hoto da yawa a cikin daƙiƙa maimakon mintuna.
Zane-zane-zane-zane-zane-zane-zane-zane-zane-zane-zane-zane-zane-zane-zane-zane-zane-zane-zane-zane-zane-zane-zane-zane-zane-zane-zane-zane-zane-zane-zane-zane-zane-zane-zane-zane-zane-zane-zane-zane-zane-zane)
Zane don tsawaita hoto sama da iyakokinsa na asali don banners ko ma'auni daban-daban.
Gyara babu abin rufe fuska, kamar canza launin kare ko sararin sama zuwa faɗuwar rana ta hanyar gyara saurin rubutu da sake canza alamun da abin ya shafa.
Hatsari & Tsare-tsare
Haƙƙoƙin hoto da yarda na iya zama haxarin doka idan ba a fayyace ba.
Ayyukan samfuri na iya bambanta a ko'ina cikin haske, ƙididdiga, da mahalli.
Ƙarya tabbataccen ƙila ba za a iya lura da shi ba sai dai idan an kula da ƙofofin amincewa.
Taswirar Hanya
Ƙayyade ma'auni na karɓa don daidaito, tunowa, da farashi na kuskure.
Gwada tare da bayanan da suka dace da ainihin yanayin samarwa.
Ƙara bita na ɗan adam don ƙarancin amincewa ko tsinkaya mai tasiri.
Bi diddigin ƙirar ƙira kuma sake ingantawa bayan canje-canjen kamara ko saitin bayanai.
Ci gaba da Bincike
Free newsletter
Get the daily AI briefing
Three verified AI stories every weekday morning, written in plain English. Free forever, no ads.
One email each weekday. Unsubscribe in one click. We never sell or share your address.
Test yourself
Take the Muse Masked Generative Imaging quiz
Instant feedback on every answer, and a shareable certificate with a verifiable ID once you pass a course.
Support free AI education. AI Understanding is a 501(c)(3) nonprofit — no ads, no paywall, ever. Make a donation
Jagora na gaba
Masked Autoencoders
Tambayoyin da ake yawan yi
What is Muse Masked Generative Imaging?
Muse samfurin rubutu-zuwa-hoto ne daga Google wanda ke samar da hotuna ta hanyar cike alamun hoto mai rufe fuska gaba daya, yana mai da shi sauri fiye da yada mataki-mataki. Yana da mahimmanci saboda ya nuna za ku iya samun ingantattun hotuna masu daidaituwa, ba tare da jinkirin ƙiyayya da yawancin janareta suka dogara da su ba.
Menene Muse ya annabta yayin tsarawa maimakon hana pixels?
Muse yana aiki a cikin sararin alamar alama, yana tsinkayar alamun hoto da aka rufe da abin token VQGAN maimakon aiki akan pixels kai tsaye.
Me yasa Muse gabaɗaya yayi sauri fiye da daidaitattun samfuran watsawa?
Muse yana cika alamomin rufe fuska da yawa a lokaci guda kuma kawai yana buƙatar ɗimbin ɗimbin zagayawa, ba kamar yawancin matakan ƙin yarda da yadawa ba.
A ina Muse ya sami fahimtar saƙon rubutu?
Ƙirƙirar yanayi na Muse akan abubuwan haɗin rubutu daga daskararre wanda aka riga aka horar da shi samfurin harshen T5-XXL, yana ba shi ƙarfin fahimtar harshe.
Menene aikin sake gyara tushen amincewa a cikin Muse?
Bayan kowane tsinkaya mai kamanceceniya, Muse yana riƙe da mafi kyawun alamu kuma yana sake rufe waɗanda ba su da ƙarfin gwiwa don daidaitawa a zagaye na gaba.
Ta yaya Muse yake sarrafa samar da mafi girman hotuna?
Muse ya fara samar da grid mai ƙarancin ƙuduri, sannan samfurin babban ƙuduri na biyu yana samar da grid mai ƙima mai ƙima.