Kedis ku xidhan hagaha · Dhexdhexaad Heerka

Kediska Jiilka Sawirka Aatooregressive

Baro sida moodooyinka sawirka autoregressive u codeeyaan sawirada sida taxanaha calaamada oo ka soo saar hal talaabo markiiba qoraalada.

Wadooyinka haga ee la xidhiidhaJiilka Sawirka Gadaashiisa
Su'aal 1 ee 8

Waa maxay macnaha 'autoregressive' ee macnaha guud ee jiilka sawirka?

Horudhac su'aal kasta8 su'aalo

Jawaabaha iyo sharraxaadaha waxay soo baxayaan markaad qaadanayso kediska.

  1. Waa maxay macnaha 'autoregressive' ee macnaha guud ee jiilka sawirka?
    • Cunsur kasta oo cusub ayaa lagu saadaaliyay in ay ku xidhan tahay curiyayaasha hore loo soo saaray
    • Cunsuri kasta oo cusub waxaa la saadaaliyaa iyadoo la isticmaalayo oo keliya qoraalka degdegga ah
    • Sawirka oo dhan halmar ayaa laga diiday qaylada aan tooska ahayn
    • Pixels ayaa la soo saaray si buuxda oo ka madax bannaan midba midka kale
  2. Sidee moodooyinka casriga casriga ah ee autoregressive sida DALL-E 1 caadi ahaan u matalaan sawir ka hor inta aan la saadaalin?
    • Sida hal vector joogto ah
    • Sida calaamad go'an oo calaamado gaar ah oo laga soo qaatay buug-codeed la bartay
    • Sida 24-bit RGB pixels cayriin oo keliya
    • Sida liiska qaababka sawirada vector
  3. Waa kuwee moodooyinka hore ee keenay sawirada hal pixel ceeriin markiiba?
    • GAN iyo VAES
    • Faafinta Deggan iyo Imagen
    • PixelRNN iyo PixelCNN
    • CLIP iyo BLIP
  4. Farsamaynkee ayaa hubiya in Transformer-ku uu kaliya tago calaamadihii hore inta lagu jiro jiilka autoregressive?
    • Ka saarida
    • Dufcaddii caadiga ahayd
    • Ka bood xidhiidhada
    • Sababta (waji-xidhan) is-fiirsashada
  5. Waa maxay cillada ugu weyn ee muunad sawireedka autoregressive?
    • Muunad-qaadiddu waa isku xig-xiga oo sidaas darteed gaabis ah
    • Laguma xidhi karo qoraal
    • Ma moodi karo faahfaahin fiican
    • Uma baahna xog tababar
  6. Farsamokee ayaa dedejisa abuurista sawirka calaamad ku salaysan iyadoo la saadaaliyay calaamado badan oo barbar socda?
    • Macalinka oo ku qasbaya
    • Nambarka isbarbar yaaca wejiyada (tusaale, MaskGIT/Muse)
    • Goynta si tartiib ah
    • Hal-kulan codayn
  7. How does the original DALL·E model use text and image tokens?
    • Separate unconnected classifiers process text and pixels
    • A diffusion decoder predicts all pixels simultaneously
    • A single autoregressive Transformer models the sequence conditioned on text and image tokens
    • The model uses only a text encoder and no image-token sequence
  8. What scale of image-text pairs did the original DALL·E paper report collecting?
    • About 250 pairs
    • About 25 thousand pairs
    • About 25 billion pairs
    • About 250 million image-text pairs