Kedis ku xidhan hagaha · Dhexdhexaad Heerka

Kediska Dhismayaasha Qoorta dhalada

Waxay hubisaa dhimista kanaalka, xisaabi kaydinta, hadhaaga rogan, cusboonaysiinta darajada hoose iyo xaddidaadda cabbirka la wadaago ee ka dambeeya caqabadaha.

Wadooyinka haga ee la xidhiidhaDhismayaasha Qoorta dhalada
Su'aal 1 ee 8

Qulqulka dhalo ee ResNet, waa maxay doorka 1x1 convolution ee ugu horreeya?

Horudhac su'aal kasta8 su'aalo

Jawaabaha iyo sharraxaadaha waxay soo baxayaan markaad qaadanayso kediska.

  1. Qulqulka dhalo ee ResNet, waa maxay doorka 1x1 convolution ee ugu horreeya?
    • Si loo yareeyo tirada kanaalada ka hor 3x3 convolution
    • Si loo kordhiyo xallinta meelaha
    • Si loo dabaqo kala-soocidda ugu dambeysa
    • Si loogu daro buuq joogto ah
  2. Waa maxay sababta dhalodu ay shabakadaha qotodheer uga dhigto mid ka jaban xisaabinta?
    • Waxay meesha ka saartaa dhammaan hawlgelinta
    • Waxay ku qabataa hawlgallo qaali ah kanaalo yar
    • Waxay isticmaashaa kernels ka weyn meel kasta
    • Waxay ka boodaa faafinta
  3. Cod-qaadaha autoencoder-ka, lakabka qoor-gooyaha ayaa ugu horreyn go'aamiya:
    • Doorashada hagaajinta
    • Qaabka faylka sawirka
    • Inta galinta la cufan
    • Cabbirka dufcadda
  4. MobileNetV2 waxay isticmaashaa kala duwanaanshiyaha fikradda qardajeexa?
    • Dhalada barkadda saafiga ah
    • Dhalada soo noqnoqda
    • Wax cilad ahi haba yaraatee
    • Qoorta haraaga ah ee leexday (ballaaran ka dibna qandaraas)
  5. Sidee LoRA ugu dabaqdaa mabaadi'da cirridka moodooyinka luqadda waaweyn?
    • Adiga oo gelinta cufan cufan ee darajada hoose (cidhiidhi ah) ee la tababari karo cusboonaysiinta miisaanka
    • Adigoo tirtiraya lakabyada dareenka
    • Adigoo labanlaabaya dhammaan miisaannada
    • Iyadoo lagu qiyaaso 1 bit
  6. Qulqulka caadiga ah ee ResNet ayaa raacaya qaabka kanaalka?
    • Balaadhan, baladhan, baladhan
    • Yaree, habraac, soo celi (tusaale, 256 ilaa 64 ilaa 256)
    • Cidhiidhi, ballaaran, cidhiidhi
    • Random halkii lakab
  7. A ResNet block reduces 256 channels to 64 before its 3×3 convolution. Ignoring the 1×1 layers, what is the 3×3 layer’s approximate multiply-add cost relative to 256 channels?
    • About one-quarter
    • About one-half
    • About one-sixteenth
    • About four times as much
  8. An autoencoder uses a narrow latent code. What constraint does this impose on reconstruction?
    • The encoder must preserve every input pixel exactly
    • The network can no longer use learned weights
    • The decoder must expand the latent code before encoding
    • The decoder must reconstruct the input from a compressed representation