Ikibazo gihuza ibibazo · Hagati Urwego

Shakisha Ikibazo

Shakisha uburyo gushakisha ibiti bikomeza amanota menshi yo mu gice hamwe nuburyo ibihano byo gutanga amanota byerekana inyandiko yanyuma.

Inzira zijyanye no kuyoboraGushakisha ibiti
Ikibazo 1 Bya 8

Niki 'ubugari bwa beam' (k) bugenzura iki mugushakisha ibiti?

Reba buri kibazo8 ibibazo

Ibisubizo nibisobanuro bigaragara mugihe ufata ikibazo.

  1. Niki 'ubugari bwa beam' (k) bugenzura iki mugushakisha ibiti?
    • Ni kangahe abakandida bakurikirana babikwa kuri buri ntambwe ya decoding
    • Uburebure ntarengwa bwurwego rwakozwe
    • Ubushyuhe bukoreshwa mubishoboka
    • Ingano yamagambo yicyitegererezo
  2. Gushakisha ibiti bitandukaniye he no kurarikira?
    • Buri gihe itoranya ikimenyetso kimwe gishoboka cyane
    • Igumana amanota menshi yo hejuru igice gikurikiranye aho kwiyemeza imwe
    • Itanga ibimenyetso ku bushake
    • Irengagiza ibimenyetso byerekana rwose
  3. Ni ukubera iki uburebure busanzwe bukoreshwa mugushakisha ibiti?
    • Kugira ibisekuru byihuse
    • Kongera ingano yamagambo
    • Kuberako guteranya log-probabilité ibogamye amanota yerekeza kumurongo mugufi
    • Kugirango wongereho ibisubizo kubisubizo
  4. Ni uwuhe murimo ushakisha ibiti cyane cyane bikwiranye?
    • Gufungura-kurangiza guhanga inkuru
    • Gukora inyandiko zitandukanye zidasanzwe
    • Gutanga ibisigo bitateganijwe
    • Ubusobanuro bwimashini, aho ibyifuzo byizerwa-bishoboka cyane
  5. Ni ubuhe buryo bubi bwo gukoresha ubugari bunini cyane bwibisekuru byafunguye?
    • Irashobora gutanga blander, gusubiramo, cyangwa gutesha agaciro inyandiko
    • Iremeza ibisohoka cyane
    • Ikuraho ibikenewe muburyo bwose bushoboka
    • Bituma icyitegererezo cyirengagiza ibyinjijwe
  6. Kuri buri ntambwe, ni gute gushakisha ibiti bihitamo abakandida bagomba gukomeza?
    • Ukurikije inyuguti zikurikirana zerekana ibimenyetso
    • Ukurikije amanota (igiteranyo cya log-probabilite), ukomeza hejuru k
    • Muguhitamo abakandida barebare
    • Guhitamo ku bushake mu bakandida bose
  7. What problem does GNMT’s coverage penalty aim to reduce in translation?
    • Repeatedly encoding the same source vocabulary
    • The use of subword units
    • Output translations that fail to cover source words
    • The number of encoder layers
  8. What does diverse beam search modify to encourage different candidate outputs?
    • It replaces the model probabilities with uniform random scores
    • It removes all but one candidate at the first step
    • It changes the training tokenizer after decoding
    • It penalizes similarity or overlap among candidates in different beam groups