Guide-yakabatana mibvunzo · Pakati Nhanho

ELECTRA Pretraining Quiz

Inotsanangura zvakatsiviwa-chiratidzo chekuonekwa, mabasa ejenareta uye kusarura, kudzidza pane zvinzvimbo zvezviratidzo, uye miganhu yekuenzanisa kuenzanisa.

Nzira dzekutungamira dzinoenderanaElectra Pretraining
Mubvunzo 1 ye 8

Nderipi basa rekudzidzira rinoshandiswa neELECTRA pachinzvimbo chekugadzira mutauro wakavharika?

Tarisa mubvunzo wese8 mibvunzo

Mhinduro uye tsananguro zvinoonekwa paunenge uchitora bvunzo.

  1. Nderipi basa rekudzidzira rinoshandiswa neELECTRA pachinzvimbo chekugadzira mutauro wakavharika?
    • Yakatsiviwa chiratidzo chekuonekwa
    • Kufanotaura kwemutsara unotevera
    • Autoregressive inotevera-izwi kufanotaura
    • Permutation mutauro wokuenzanisira
  2. Sei ELECTRA ichinyanya kuita compute-inoshanda kupfuura BERT?
    • Inoshandisa mazwi mashoma zvikuru
    • Iyo inoverengera chiratidzo chekudzidza pamusoro pese tokeni, kwete chete akafukidzwa
    • Inongodzidzisa pane yakavharwa 15% yezviratidzo
    • Inodarika iyo yekutarisisa meshini zvachose
  3. Ndeipi basa rinoitwa nediki jenareta network muELECTRA?
    • Inoisa zvibodzwa zvekupedzisira nemhando
    • Inomanikidza modhi mushure mekudzidziswa
    • Iyo inogadzira zvinogoneka zvinotsiva tokeni kuti anosarura atonge
    • Inoturikira zvinyorwa pakati pemitauro
  4. Chii chinoitika kune jenareta mushure mekunge pretraining yapera?
    • Yakanyatsogadzirirwa basa rezasi
    • Inosanganiswa nerusarura
    • Inova decoder yechizvarwa
    • Inoraswa; kusarura chete kunoshandiswa
  5. ELECTRA yakagadziridzwa zvakanyanya nevaongorori kubva kumasangano api?
    • Google uye Stanford
    • Meta AI neNYU
    • OpenAI uye Microsoft
    • DeepMind uye Oxford
  6. Ndeipi yakazotevera modhi yakawedzera ELECTRA's yakatsiviwa-chiratidzo-yekuona pfungwa?
    • GPT-2
    • DeBERTa-v3
    • T5
    • Word2Vec
  7. What happens if ELECTRA’s generator proposes the original token at a position?
    • It is always labeled replaced because the generator proposed it
    • The position is removed from the input
    • That token is treated as original by the discriminator’s replaced-token label
    • The whole sequence is discarded
  8. When comparing ELECTRA with another pretraining method, what makes a compute claim meaningful?
    • Compare model names alone
    • Assume every ELECTRA run uses one fixed compute ratio
    • Ignore generator training cost
    • Specify model size, data, training budget, and evaluation setting