Ikibazo gihuza ibibazo · Hagati Urwego

CLIP na Icyerekezo-Ururimi Model Ikibazo

Ongera usubiremo CLIP isangiwe hamwe no gutandukanya, hamwe na zeru-zeru zeru, imiryango ya encoder yumwimerere, kubara amanota-amanota, hamwe na UNCLIP.

Ikibazo 1 Bya 8

Niki gitekerezo cyibanze inyuma ya CLIP?

Reba buri kibazo8 ibibazo

Ibisubizo nibisobanuro bigaragara mugihe ufata ikibazo.

  1. Niki gitekerezo cyibanze inyuma ya CLIP?
    • Gucomeka amashusho muri dosiye nto
    • Gushyira amashusho hamwe ninyandiko mumwanya umwe usangiwe gushira
    • Gukora moderi ya 3D uhereye kumafoto
    • Guhindura imvugo mu nyandiko
  2. Ni ubuhe buryo bwo guhugura CLIP ikoresha?
    • Kwiga gushimangira ibitekerezo byabantu
    • Kuranga intoki buri kintu muri buri shusho
    • Kwiga gutandukanya gukurura ishusho-inyandiko byombi hamwe
    • Gukoporora ibipimo bivuye muburyo bwo kuvuga
  3. What does zero-shot classification mean in the original CLIP transfer setup?
    • Using text-defined candidate categories without new task-specific supervised classifier fitting
    • Using a model that never received any pretraining data
    • Running without any computation or electricity
    • A guarantee that every category is classified correctly
  4. Nigute CLIP ipima niba ishusho nibisobanuro bihuye?
    • Kugereranya ingano ya dosiye
    • Kubara pigiseli ya buri bara
    • Mugenzura igihe cyerekana ishusho
    • Kubijyanye na cosine ihuza ibice byabo
  5. Hafi yamakuru angahe CLIP yatojwe?
    • Ibihumbi bike byanditseho amaboko
    • Gusa inyandiko, nta mashusho
    • Hafi ya miliyoni 400 zishusho-inyandiko zivuye kurubuga
    • Icyegeranyo kimwe cyatunganijwe
  6. How does the cited unCLIP image-generation paper use CLIP representations?
    • The original standalone CLIP encoder directly outputs every image pixel
    • A prior supplies CLIP image embeddings to a generative decoder
    • It removes all image representations and uses only filenames
    • It uses CLIP only as a network-encryption protocol
  7. Which image-encoder families were evaluated in the original CLIP paper?
    • Only recurrent text encoders
    • Only a mandatory linear pixel average
    • Only a raw audio codec
    • Modified ResNets and Vision Transformers
  8. A CLIP batch contains 32 paired images and texts. How many entries are in its full cross-modal score matrix?
    • 1024
    • 32
    • 64
    • 992