quiz bu lëkkale ak tegtal · Dëgër Niveau

HuBERT Quiz buy saytu sa bopp

Saytu ni HuBERT di defaree ay mbir yu bawoo ci cluster ngir audio buñ maskeer, baamtu etiket yi, ak jàngat misaal yiñ jàng ci benchmark wax.

Yooni njiit yu jëm ciKaddu bu Hubert di saytu boppam
Laaj 1 de 8

naka la HuBERT di defaree ay mébet yu mu jéema seetlu ci diiru tàggat yaram?

Xoolaatal laaj bu nekk8 laaj

Sooy def quiz bi, tontu yi ak leeral yi dañuy feeñ.

  1. naka la HuBERT di defaree ay mébet yu mu jéema seetlu ci diiru tàggat yaram?
    • Offline k-dafay tekki boole ay man-mani audio ci ay yunit yu nëbbu
    • Transkripsioŋ yu nit
    • Etiketu bruit yu bari
    • Contrast ak misaali distractor
  2. Ban xeetu bind bu siiw moo inspiré HuBERT ci tàggat yaram bu maskewu?
    • GPT-3
    • Bert
    • Resnet
    • Word2Vec
  3. naka lay def ba HuBERT gëna suqali ay etiketam ci wàllu tàggat yaram yu toppalante?
    • Dafay jël yeneen annotatër
    • Dafay yokk bruit bu metti
    • Dafay boolewaat ay cluster di jëfandikoo model bi boppam gëna suqali ci biir
    • Dafay gelé clustering bu njëkk bi ba fàww
  4. Lan mooy màndarga yiñ gëna boole ci iteration bu njëkk bi ci HuBERT?
    • Baatu transkripsioŋ yu mujj
    • Pixel nataal
    • ID kàddukat
    • Man-mani akustik yu yomb yu MFCC
  5. Lan moo waral HuBERT mëna jàng structure bu am njariñ doonte clustering bi jaarul yoon?
    • Dafay soxla ludul seetlu ID cluster yi ci barab yuñ maske, te loolu dafay wéy di nekk liggéey buy wéy
    • Dafay baña bàyyi xel ci cluster yi
    • Dafay jëfandikoo ab clustering oracle bu jaar yoon
    • Dafay jàng set test bi ci boppam
  6. Which pretraining challenge does HuBERT address by using offline cluster labels?
    • Speech has no temporal structure
    • There is no lexicon of sound units available during pretraining
    • Every speaker uses the same pitch
    • Audio contains no repeated patterns
  7. In the HuBERT paper, what relative WER reductions did the 1B-parameter model report on dev-other and test-other?
    • Exactly 50% on both sets
    • No measurable change on either set
    • Up to 19% on dev-other and 13% on test-other
    • Up to 13% on dev-other and 19% on test-other
  8. Which benchmark sets did the original HuBERT paper use for speech-recognition results?
    • NataalNet ak COCO
    • GLUE and SuperGLUE
    • SQuAD and TriviaQA
    • LibriSpeech and Libri-light