Kedis ku xidhan hagaha · Dhexdhexaad Heerka

RNN-Transducer Model Kediska

Faham sida moodooyinka RNN-Transducer ay isugu daraan fiimyada dhawaaqa iyo taariikhda soo-saarka ee baahinta hawlaha qorista ee taxanaha ah.

Wadooyinka haga ee la xidhiidhaMoodooyinka Rnn Transducer
Su'aal 1 ee 8

Waa kuwee xaddidaad CTC ah ayuu RNN-Transducer si gaar ah uga hadlayaa?

Horudhac su'aal kasta8 su'aalo

Jawaabaha iyo sharraxaadaha waxay soo baxayaan markaad qaadanayso kediska.

  1. Waa kuwee xaddidaad CTC ah ayuu RNN-Transducer si gaar ah uga hadlayaa?
    • Malaha madax-bannaanida shuruudaysan ee CTC, taas oo ka hortagaysa qaabaynta ku-tiirsanaanta wax-soo-saarka
    • CTC gabi ahaanba ma farsamayn karto maqalka
    • CTC waxay u baahan tahay GPU-yada si ay u shaqeeyaan
    • CTC waxay u shaqeysaa sawirada kaliya
  2. Waa maxay saddexda qaybood ee ugu muhiimsan ee RNN-Transducer?
    • Encoder, codeer, iyo takooriye
    • Encoder (Shabakad qoraal ah), Shabakadda saadaalinta, iyo shabakad wadaag ah
    • Tokenizer, Embedder, iyo softmax kaliya
    • Laba madax dareenka iyo router ah
  3. Doorkee ayay ka ciyaartaa shabakada saadaasha ee RNN-T?
    • Waxa ay cadaadinaysaa qaab-dhismeedka codka
    • Waxay cabbirtaa mugga makarafoonka
    • Waxay u shaqeysaa sida qaab luqadeed, oo ku qaboojinaya calaamadihii hore loo sii daayay
    • Waxay doorataa heerka saamiga
  4. Maxay RNN-T u taageertaa qulqulka daahsoonaanta hoose si dabiici ah?
    • Waxa ay socodsiisaa dhammaan hadalka hal mar ka hor inta aanay wax sii dayn
    • Waxay iska indha tirtaa nidaamka waqtiga oo dhan
    • Waxa kaliya oo ay ku shaqeysaa faylal horay loo duubay
    • Isku toosinteedu waa monotonic iyo bidix-ilaa-midig, iyada oo ku sii socota maqalka oo bannaan
  5. Decodeynta RNN-T, maxay samaynaysaa soo saarista calaamad madhan?
    • Horumarka talaabada xigta ee wakhtiga codka
    • Dhammaysta qoraalka oo dhan
    • Wuxuu tirtiraa kelmadii hore
    • U beddelaa luqad kale
  6. Waa kuwee nooca codeeyaha ee sii kordhaya ee beddelaya LSTM-yada wax soo saarka casriga ah ee nidaamyada RNN-T?
    • Go'aan geedaha
    • Codeeyayaasha Conformer
    • Boorsada-ereyada cad
    • Qaababka qarsoon ee Markov
  7. What two histories condition the RNN-T joint network’s output distribution?
    • Only the future transcript and a speaker photo
    • Only the last acoustic frame, without output context
    • The encoder representation of acoustic input and the prediction-network representation of prior output labels
    • The optimizer state and learning rate
  8. What alignment behavior does an RNN-T allow before advancing to the next audio frame?
    • It must emit exactly one label for every frame
    • It advances time only after a word-final token
    • It generates all output before reading any audio
    • It can emit zero or more nonblank labels at the current frame, then a blank advances time