InoteveraGaidhi rinotevera
Kupa Mavara muAI-Yakagadzirwa Mifananidzo
AI Yekutarisa
Gwara reVisual AI
Imagen Chinyorwa-ku-Mufananidzo
Imagen ndeye Google's text-to-image system inoshandura tsananguro dzakanyorwa kuita mifananidzo yemifananidzo.
Teerera kugwaro iriAI-yakarondedzerwa · ~2 min
Pa peji ino2 min verenga
Pfupiso
Its headline finding was that a large frozen language model, not a bigger image network, was the biggest driver of quality.
Kudzika Kwakadzika
Yakaziviswa ne Google Tsvagiridzo muna 2022, Imagen yakaratidza kuti kunzwisisa zvakanyanya kukurumidza kunokosha sekudhirowa zvakanaka. Panzvimbo peiyo CLIP-maitiro ekunyora encoder, Imagen inoshandisa hombe yakadzidziswa mameseji encoder (T5-XXL) inochengetwa yakaomeswa nechando, yobva yadyisa iwo akapfuma mitauro yakamisikidzwa kuita modhi yekuparadzira. Iyo inogadzira diki 64x64 mufananidzo uye inoshandisa maviri epamusoro-resolution diffusion nhanho kusvika kumusoro kusvika 1024x1024. Chikwata chakaunzawo 'dynamic thresholding' kuitira kuti mavara arambe akatsiga pakutungamirwa kwepamusoro, uye akavaka DrawBench, mucherechedzo wekutemesa kuyedza kuverenga, hukama hwenzvimbo, uye misanganiswa isingawanzo. Shanduro dzakazotevera, Imagen 2 uye Imagen 3, zvakarodzwa, zvinyorwa, uye nekukurumidza kuvimbika, uye ikozvino shandisa Google's mifananidzo maturusi.
Kuitwa muNyika yeChokwadi
Kugadzira zvinoonekwa zvekushambadzira zvepikicha kubva muchidimbu chakanyorwa pasina kupfura kwepikicha
Kugadzira mifananidzo yepfungwa yenganonyorwa kana mabhuku evana kubva mumitsara inotsanangura
Kugadzira zvigadzirwa mockups uye zviitiko zvakasiyana zvee-commerce zvinyorwa
Kuona pfungwa dzesainzi kana dzedzidzo, seshanduro yenyanzvi inotsanangurwa nemutauro wakajeka
Ramangwana reImagen Chinyorwa-ku-Mufananidzo
Mutsara weImagen uri kufamba uchienda kune zvirinani kupa mavara mukati memifananidzo, kukurumidza kukurumidza kutevera kune zvakaomarara, uye nekukurumidza sampling. Tarisira kusanganiswa kwakadzama nemhando dzemitauro kuitira kuti sisitimu 'ikonzere' nezve chikumbiro isati yadhirowa, pamwe neakasimba watermarking seSynthID yepakutanga. Sezvo ichibatanidza Google's zvigadzirwa uye Gemini ecosystem, chinotarisisa chinochinjira kune chakavimbika, chakachengeteka, chinodzoreka chizvarwa kwete chitsva.
Kuzvishandisa
Nei zvichikosha
- Kumhanya uye chiyero. Visual AI inogona kuita otomatiki yekuongorora, yekuona, uye yekumaka mabasa pachiyero.
- Vaka sarudzo. Zvikwata zvekugadzira zvinogona prototype pfungwa nekukurumidza nekudzokororwa kwemaoko mashoma.
- Team uye workflow. Mashandisirwo anogona kushandisa masaini emifananidzo nemavhidhiyo ayo aimbove akaoma kugadzirisa.
Chenjera
- Kodzero dzemifananidzo uye kubvumirwa kunogona kuve njodzi dzepamutemo kana provenance isina kujeka.
- Kuita kwemuenzaniso kunogona kusiyanisa kupenya, huwandu hwevanhu, uye nharaunda.
- Manyepo enhema anogona kusacherechedzwa kunze kwekunge zvikumbaridzo zvekuvimba zvikatariswa.
Matanho anotevera
- Tsanangura maitiro ekugamuchirwa echokwadi, kurangarira, uye mutengo wekukanganisa.
- Edzai nedata rinoenderana nemamiriro chaiwo ekugadzira.
- Wedzera ongororo yemunhu kune yakaderera-kusavimbika kana yakakwirira-inokanganisa kufanotaura.
- Tevera modhi kudonha uye simbisa mushure mekuchinja kwekamera kana dataset.
Ramba Uchiongorora
Tsamba yemahara
Tora iyo yemazuva ese AI pfupiso
Nhatu dzakasimbiswa dzeAI nyaya mangwanani ega ega pakati pevhiki, dzakanyorwa neChirungu chakajeka. Yemahara zvachose, hapana ads.
Imwe email zuva rega rega revhiki. Bvisa kunyoresa kamwe chete. Hatimbotengese kana kugovera kero yako.
Zviedze iwe pachako
Take the Imagen Text-to-Image quiz
Mhinduro yepakarepo pamhinduro yega yega, uye chitupa chinogovaniswa chine ID inobatika kana wapedza kosi.
Tsigira dzidzo yemahara yeAI. AI Understanding ndeye 501(c)(3) isingabatsiri - hapana ads, hapana paywall, zvachose. Ita chipo
Mibvunzo inowanzo bvunzwa
What is Imagen Text-to-Image?
Imagen ndeye Google's text-to-image system inoshandura tsananguro dzakanyorwa kuita mifananidzo yemifananidzo. Musoro waro wekutsvaga waive wekuti modhi yakakura yemutauro wakaoma nechando, kwete network yemifananidzo yakakura, yaive mutyairi mukuru wemhando.
In the original Imagen study's scaling comparison, which change produced larger image-quality and text-alignment gains?
The reported comparisons favored scaling the text encoder; this is a study result, not a universal rule for every architecture.
Ndeipi encoder yemavara inovimbwa nayo Imagen?
Imagen inoshandisa hombe, zvinyorwa-chete pretrained T5-XXL encoder, inochengetwa nechando panguva yekudzidziswa.
Imagen inosvika sei pakugadziriswa kwepamusoro?
Iyo inogadzira 64x64 mufananidzo wozokwira kumusoro kuburikidza nepamusoro-resolution diffusion modhi kusvika 1024x1024.
What problem did original Imagen's dynamic thresholding address under strong guidance?
The method clips and rescales predicted pixels to limit saturation under large guidance weights.
Chii chinonzi DrawBench?
DrawBench ibhenji Google yakaunzwa neImagen kuverengera kuverenga, hukama hwepakati, uye zvisingawanzo kukurumidza kukurumidza.
Ramba uchidzidza
Mazano anoenderana
Mamwe madhairekitori akasarudzirwa nyaya iyi