Maswali yanayounganishwa na mwongozo · Kati Kiwango

Maswali ya Wav2Vec 2.0

Jaribio la kichocheo cha kujifunzia mapema cha Wav2Vec 2.0, lengo linganishi, na vikomo vya matokeo yake ya LibriSpeech yenye lebo ya chini.

Njia za mwongozo zinazohusianaWav2vec 2 0
Swali 1 ya 8

Ni tatizo gani kuu katika utambuzi wa usemi ambalo Wav2Vec 2.0 iliundwa kushughulikia?

Hakiki kila swali8 maswali

Majibu na maelezo huonekana unapojibu maswali.

  1. Ni tatizo gani kuu katika utambuzi wa usemi ambalo Wav2Vec 2.0 iliundwa kushughulikia?
    • Uhaba wa sauti yenye lebo (iliyonakiliwa).
    • Faili za sauti zikiwa kubwa mno kuhifadhi
    • Maikrofoni kuwa ghali sana
    • Maandishi ni magumu kuweka tokeni
  2. Je, ni aina gani ya lengo la kujifunza ambalo Wav2Vec 2.0 hutumia wakati wa mafunzo ya awali?
    • Uainishaji wa fonimu moja kwa moja na lebo kamili
    • Hasara tofauti dhidi ya sehemu zilizofichwa zilizofichwa
    • Uundaji upya wa pixel
    • Mafunzo ya kuimarisha kutoka kwa maoni ya kibinadamu
  3. Ni sehemu gani ya kwanza huchakata muundo mbichi wa wimbi katika Wav2Vec 2.0?
    • LSTM ya kawaida
    • Sampuli ya kueneza
    • Kisimbaji cha kipengele cha CNN cha safu nyingi
    • Kichimbaji cha MFCC kilichotengenezwa kwa mikono
  4. In a reported LibriSpeech experiment, what small labeled-data setup demonstrated low-label potential?
    • 10 minutes unlabeled with no pretraining
    • 53,000 hours labeled with no pretraining
    • One hour labeled with no unlabeled data
    • 10 minutes labeled after pretraining on 53,000 hours of unlabeled speech
  5. Je, kitabu cha msimbo kilichojifunza kina jukumu gani katika Wav2Vec 2.0?
    • Inatofautisha sauti inayoendelea kuwa vitengo vya usemi vyenye kikomo
    • Huhifadhi nywila za mtumiaji
    • Inabana nakala ya mwisho
    • Inachagua faida ya maikrofoni
  6. Ni modeli gani kubwa ya lugha nyingi imejengwa kwa mbinu ya Wav2Vec 2.0?
    • GPT-2
    • XLS-R
    • Stable Diffusion
    • ResNet-50
  7. What does the Transformer contribute during Wav2Vec 2.0 pretraining?
    • Converts transcripts into audio
    • Clusters final text into language labels
    • Contextualizes latent speech features around masked spans
    • Removes the feature encoder
  8. What is typically done after self-supervised pretraining to create an ASR model?
    • Replace the audio encoder with an image classifier
    • Train only on unlabeled text
    • Discard the model and use k-means alone
    • Fine-tune with transcribed speech