Maswali ya Usanifu wa Conformer

Majibu na maelezo huonekana unapojibu maswali.

1. Je, usanifu wa Conformer unachanganya njia gani mbili kwenye block moja?
   • Convolution na tahadhari binafsi
   • Kujirudia na kuunganisha
   • Miti ya maamuzi na nguzo
   • Utafutaji wa boriti na tokeni
2. Kwa nini kuchanganya convolution na tahadhari ni muhimu hasa kwa hotuba?
   • Umakini hushughulikia maelezo ya fonimu ya ndani huku ubadilishanaji hushughulikia muktadha wa kimataifa
   • Convolution inanasa mifumo ya ndani huku umakinifu unanasa muktadha wa masafa marefu
   • Zote mbili ni muhimu kwa picha, sio sauti
   • Wala haisaidii; ni kwa kasi tu
3. 'Macaron' au muundo wa sandwich wa block ya Conformer ni nini?
   • Tabaka mbili za umakini karibu na msongamano mmoja
   • Safu moja ya kusambaza mbele pekee
   • Uangalifu na ubadilishaji umefungwa kati ya moduli mbili za nusu-hatua za kusonga mbele
   • Convolution iliyopangwa juu ya convolution
4. Ni shirika gani lilianzisha Conformer, na katika mwaka gani?
   • OpenAI mwaka wa 2018
   • Meta mwaka wa 2015
   • Microsoft mwaka wa 2012
   • Google mwaka wa 2020
5. Moduli ya ubadilishaji ndani ya Conformer kawaida inajumuisha nini?
   • Ushawishi wa uhakika, GLU, ushawishi wa kina, urekebishaji, na kuwezesha Swish
   • Safu moja tu iliyounganishwa kikamilifu
   • Seli ya LSTM inayojirudia
   • Kiainishi cha msitu nasibu
6. On LibriSpeech test-clean/test-other, what WER did the original Conformer paper report with an external language model?
   • 2.1% / 4.3%
   • 1.9% / 3.9%
   • 9.1% / 9.3%
   • 0.1% / 0.3%
7. What positional information does the Conformer paper add to attention?
   • A geographic GPS coordinate for every speech frame
   • A fixed class label for each utterance
   • Relative positional encoding
   • A separate video frame index
8. What word error rates did the Conformer paper report without an external language model on LibriSpeech test/test-other?
   • 20.1% / 40.3%
   • 0.1% / 0.3%
   • 12.1% / 14.3%
   • 2.1% / 4.3%