Idanwo ti o sopọ mọ itọsọna · Rọrun Ipele

Ni àídájú GANs adanwo

Checks how conditional inputs guide both GAN networks, how discriminators assess matched pairs, and how conditioning can be injected.

Jẹmọ itọsọna awọn ọnaGANs ni àídájú
Ibeere 1 ti 8

Kini iyatọ akọkọ laarin GAN ni majemu ati GAN boṣewa kan?

Awotẹlẹ gbogbo ibeere8 ibeere

Awọn idahun ati awọn alaye han bi o ṣe n mu ibeere naa.

  1. Kini iyatọ akọkọ laarin GAN ni majemu ati GAN boṣewa kan?
    • O ṣe ifunni alaye afikun bi aami kilasi sinu awọn nẹtiwọọki mejeeji lati ṣakoso iṣelọpọ
    • O nlo meji Generators
    • O yọ awọn discriminator kuro
    • O ṣiṣẹ nikan lori ọrọ, kii ṣe awọn aworan
  2. Ninu cGAN ti o gba ikẹkọ lori awọn nọmba ti o ni aami, kini o le ṣe ti GAN lasan ko le?
    • Ṣiṣe laisi igbewọle ariwo eyikeyi
    • Beere nọmba kan pato gẹgẹbi '7'
    • Foju iyasoto patapata
    • Ṣe ina dudu-ati-funfun ariwo nikan
  3. Kini gbọdọ ṣayẹwo iyasoto cGAN, kọja boya aworan kan dabi gidi?
    • Ọna faili ti aworan naa
    • Ipinnu ni megapixels
    • Boya aworan naa baamu aami ti a fun ni
    • Oṣuwọn ikẹkọ ikẹkọ
  4. Ewo ni ọna ti o wọpọ lati fi ranse ifihan agbara agbara si monomono?
    • Pipaarẹ fekito ariwo
    • Ikẹkọ laisi eyikeyi aami
    • Lilo sensọ ti ara ọtọtọ
    • Sisọpọ aami si igbewọle ariwo
  5. Awọn GAN ti o wa ni ipo ti gbe ipilẹ le fun agbara nigbamii?
    • Ọrọ-si-aworan ati aworan-si-image iran
    • Yiyara CPUs
    • Sisẹ imeeli spam
    • Atọka aaye data
  6. Ninu awọn awoṣe itankale ode oni, imọran cGAN ti kondisona nigbagbogbo han bi ẹrọ wo?
    • Aileto àdánù clipping
    • Agbelebu-akiyesi on a ọrọ tọ
    • Kikun aworan afọwọṣe nikan
    • Pa iṣeto ariwo kuro
  7. Why must a conditional discriminator see both the generated image and its conditioning label?
    • It uses the label to choose the learning-rate schedule
    • It converts the label into a random noise seed only
    • It needs to judge whether the image is realistic and whether it matches the requested condition
    • It prevents the generator from receiving noise
  8. Besides concatenating a class label to the noise vector, which conditioning method is described in the guide?
    • A spectral-radius constraint
    • An inverse Hessian update
    • A sequence-level BLEU reward
    • A projection layer that scores compatibility between label and image features