Guide-yakabatana mibvunzo · Pakati Nhanho

Contrastive Decoding Quiz

Miedzo nyanzvi-amateur zvibodzwa, yakagovaniswa-chiratidzo kudzvinyirirwa, plausibility kusefa, decoding-nguva kushandiswa, uye kuongororwa kwechizvarwa chakasiyana.

Nzira dzekutungamira dzinoenderanaContrastive Decoding
Mubvunzo 1 ye 8

Mukududzira kwakasiyana, ibasa rei rinoitwa nemuenzaniso mudiki we'amateur'?

Tarisa mubvunzo wese8 mibvunzo

Mhinduro uye tsananguro zvinoonekwa paunenge uchitora bvunzo.

  1. Mukududzira kwakasiyana, ibasa rei rinoitwa nemuenzaniso mudiki we'amateur'?
    • Yayo log-mikana inobviswa kudzvinyirira yakagovaniswa kutadza modes
    • Inogadzira chinyorwa chekupedzisira chinobuda zvakananga
    • Yakanyatsogadziriswa pazvikanganiso zvenyanzvi
    • Inotsiva nyanzvi kana nyanzvi isina chokwadi
  2. Sei kubvisa zvingangoitika zveaamateur zvichideredza kudzokorora uye kuburitsa bland?
    • Mudzidzisi akadzidziswa kusambodzokorora mazwi
    • MaGeneric uye anodzokororwa tokeni anofarirwa nemhando mbiri, saka musiyano unodzima
    • Iye amateur anodzima tokeni dzakadzokororwa mushure mechizvarwa
    • Kudzokorora hazvibviri kana mhando mbiri dzavhota
  3. What does the adaptive plausibility constraint do in contrastive decoding?
    • It guarantees every remaining continuation is coherent and true
    • It replaces the expert distribution with the amateur distribution
    • It filters out candidates whose expert probability is too low relative to the expert’s top token
    • It forces every token to have equal probability
  4. Zvingangove kuti chii chinonzi scoring formula mukusiyanisa decoding?
    • log p_amateur minus log p_nyanzvi
    • p_nyanzvi nguva p_amateur
    • Avhareji yezviviri zvingangoitika
    • log p_nyanzvi kubvisa coefficient nguva log p_amateur
  5. Yakakosha inoshanda mukana wekusiyanisa decoding ndeyekuti:
    • Inoshanda panguva yekufungidzira pasina kudzidziswa kwakawedzerwa
    • Inoda kudzidzirazve mamodheru maviri pamwe chete
    • Inoshanda chete pazvikamu zvemabasa
    • Inobvisa kudiwa kwemuenzaniso mukuru
  6. Ndeipi nzira inotevera inotambanudzira pfungwa yekusiyanisa kune imwe modhi yezvikamu zvayo?
    • Beam search
    • DoLa, iyo inosiyanisa ekutanga neanonoka maseru
    • Kukara decoding
    • Kuwedzera tembiricha
  7. The expert assigns probability 0.4 to its top token and the plausibility threshold is 0.1 of that value. What is the minimum probability for a candidate to remain eligible?
    • 0.4
    • 0.1
    • 0.04
    • 0.004
  8. Why can contrastive decoding require more inference work than using the expert model alone?
    • It must retrain both models for every prompt
    • It replaces the vocabulary with image patches
    • It runs only the amateur model
    • It scores candidates with both an expert and an amateur model at inference time