Ikibazo gihuza ibibazo · Biragoye Urwego

Ikibazo gikaze-Kumenya Kugabanya Ikibazo

Kugenzura ivugururwa ryibice bibiri bya SAM, radiyo yabaturanyi, igipimo cy-uburemere bwa gradient hamwe nubucuruzi hagati yuburinganire na compte.

Inzira zijyanye no kuyoboraGukarishye Kumenya Kugabanuka
Ikibazo 1 Bya 8

Ni ubuhe bwoko bwa SAM igerageza kubona?

Reba buri kibazo8 ibibazo

Ibisubizo nibisobanuro bigaragara mugihe ufata ikibazo.

  1. Ni ubuhe bwoko bwa SAM igerageza kubona?
    • Nibura byibuze aho igihombo kiguma hasi murwego rwuburemere
    • Nibura cyane bishoboka
    • Nibura hamwe nibipimo byinshi
    • Nibura byibuze bigerwaho murwego rumwe
  2. Nibangahe imbere-gusubira inyuma intambwe isanzwe ya SAM isaba?
    • Kimwe cya kabiri
    • Hafi ya bibiri
    • Nukuri
    • Icumi cyangwa byinshi
  3. Niki radiyo hyperparameter rho igenzura muri SAM?
    • Igipimo cyo kwiga cyangirika
    • Umubare wibihe byamahugurwa
    • Ingano yabaturanyi ikikije ibiro birinzwe kurinda igihombo
    • Birashoboka ko bata ishuri
  4. Niyihe yambere mu ntambwe ebyiri za SAM kuri buri itera?
    • Gabanya icyitegererezo
    • Gereranya uburemere bwintambwe zose zashize
    • Ongera usubize leta
    • Fata intambwe yo kuzamuka yubunini rho werekeza kubintu bibi cyane hafi
  5. Kuki SAM ikunze kunoza imbaraga zo kuranga urusaku?
    • Kuberako minima iringaniye idakunze gufata mu mutwe ibirango byangiritse
    • Kuberako isiba ibirango bisakuza byikora
    • Kuberako ikoresha amagambo manini
    • Kuberako yikubye inshuro eshatu ingano
  6. Ninde muribi ni variant yagenewe kugabanya igiciro cya compte ya SAM?
    • ReLU
    • ESAM (SAM ikora neza)
    • BatchNorm
    • AdaGrad
  7. In SAM’s second pass, which gradient is used to update the original model weights?
    • The gradient of a random label
    • The first pass gradient without evaluating the perturbed point
    • The gradient evaluated at the perturbed nearby weights
    • A gradient from a different model checkpoint
  8. If SAM’s neighborhood radius rho is set to zero, what happens to its perturbation step?
    • The model parameters are all set to zero
    • The loss becomes independent of the weights
    • The optimizer takes an infinite step
    • There is no weight perturbation, so the neighborhood reduces to the current point