Kediska Aadan iyo La qabsiga Hagaajinta

Jawaabaha iyo sharraxaadaha waxay soo baxayaan markaad qaadanayso kediska.

1. Waa maxay labada qiyaasood ee uu Aadan ku raad-joogo halbeeg kasta?
   • Celceliska orodka gradients iyo jaangooyooyinka labajibbaaran
   • Heerka ugu sarreeya iyo kii ugu yaraa ee abid la arkay
   • Heerka waxbarashada iyo cabbirka dufcadda
   • Qiimaha miisaanka iyo calaamadiisa
2. Muxuu Aadam ugu dabaqayaa sixitaanka eexda qiyaasihiisa xilligan?
   • Sababtoo ah gradients had iyo jeer waa taban goor hore tababarka
   • Sababtoo ah celceliska dhaqdhaqaaqa wuxuu ka bilaabmaa eber waxayna u eexanayaan eber marka hore
   • Si heerka waxbarashadu u weynaado muddo ka dib
   • Si aad uga saartid dhibcaha xogta dufcada
3. Waa maxay faraqa ugu weyn ee u dhexeeya Aadan iyo AdamW?
   • AdamW wax dardar ah ma isticmaalo
   • AdamW waxa uu meesha ka saaray qiyaasta daqiiqada labaad
   • AdamW waxa uu curyaamiyaa qudhunka miisaanka cusboonaysiinta ku salaysan gradient
   • AdamW waxa uu ka shaqeeyaa shabakadaha kacaanka
4. Doorkee ayuu ereyga yar ee epsilon ka ciyaartaa qaanuunka cusboonaysiinta Aadan?
   • Waxay dejinaysaa xoogga dhaqdhaqaaqa
   • Waxay xakameysaa tirada xilliyada tababarka
   • Waxay kordhisaa xajmiga gradient-ka
   • Waxay ka hortagtaa kala qaybinta eber waxayna dejisaa cusbooneysiinta
5. Waa maxay sababta Aadan inta badan loogu tilmaamo 'laqabsi'?
   • Waxay si wax ku ool ah u siinaysaa halbeeg kasta cabbirkeeda cabbirka ah
   • Waxay beddeshaa qaab-dhismeedka shabakadda inta lagu jiro tababarka
   • Waxay la qabsataa xogta xogta qaabka
   • Waxay u kala beddeshaa kala soocida iyo dib u noqoshada si toos ah
6. Which Adam parameter controls the exponential decay rate of its squared-gradient estimate?
   • Beta1
   • Beta2
   • Epsilon
   • The weight-decay coefficient
7. A parameter repeatedly receives large gradients. Which Adam quantity helps scale down its effective step?
   • The number of model layers
   • The batch-normalization mean
   • The second-moment estimate of squared gradients
   • The AdamW weight-decay term alone
8. What happens when beta2 is set closer to one, all else equal?
   • The optimizer stops using gradients
   • The first-moment estimate is deleted
   • Weight decay becomes decoupled automatically
   • The squared-gradient estimate decays more slowly and retains a longer history