Guide-yakabatana mibvunzo · Pakati Nhanho

Autoregressive Image Generation Quiz

Learn how autoregressive image models encode images as token sequences and generate them one step at a time from text prompts.

Nzira dzekutungamira dzinoenderanaAutoregressive Image Generation
Mubvunzo 1 ye 8

Chii chinonzi 'autoregressive' chinorevei mumamiriro ekugadzirwa kwemifananidzo?

Tarisa mubvunzo wese8 mibvunzo

Mhinduro uye tsananguro zvinoonekwa paunenge uchitora bvunzo.

  1. Chii chinonzi 'autoregressive' chinorevei mumamiriro ekugadzirwa kwemifananidzo?
    • Chinhu chitsva chega chega chinofanotaurwa chakamisikidzwa pazvinhu zvakambogadzirwa
    • Chinhu chitsva chega chega chinofanotaurwa uchishandisa chete chinyorwa chekukurumidza
    • Mufananidzo wacho wose unodonhedzwa kamwechete kubva kune ruzha rwakasiyana
    • Pixels anogadzirwa zvachose akazvimirira kubva kune mumwe
  2. Ko mamodheru emazuva ano anoregressive mafananidzo seDALL-E 1 anowanzo miririra mufananidzo usati wazvifembera?
    • Sevheta imwe chete inoenderera
    • Segridi yakagadziriswa ye discrete tokens kubva kune yakadzidziswa codebook
    • Sezvo mbishi 24-bit RGB pixels chete
    • Serondedzero yevector graphics maumbirwo
  3. Ndeapi mamodheru ekutanga akagadzira mifananidzo imwe mbishi pixel panguva?
    • GANs uye VAEs
    • Yakagadzikana Diffusion uye Imagen
    • PixelRNN uye PixelCNN
    • CLIP uye BLIP
  4. Ndeipi nzira inovimbisa kuti Transformer inongotarisa kune ekutanga tokens panguva ye autoregressive chizvarwa?
    • Dropout
    • Batch normalization
    • Skip links
    • Causal (masked) kuzvidzora
  5. Ndeipi huru inoshanda dhizaini yeautoregressive image sampling?
    • Sampling inotevedzana uye saka inononoka
    • Haikwanise kumiswa pamavara
    • Haikwanise kuenzanisira zvakadzama
    • Izvo zvinoda hapana data yekudzidziswa
  6. Ndeipi tekinoroji inomhanyisa kugadzirwa-kwakavakirwa mufananidzo chizvarwa nekufanotaura akawanda matokeni akafanana?
    • Kumanikidza mudzidzisi
    • Masked parallel decoding (e.g., MaskGIT/Muse)
    • Gradient clipping
    • Imwe-inopisa encoding
  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