Su'aalaha Encoder-Decoder Architectures

Jawaabaha iyo sharraxaadaha waxay soo baxayaan markaad qaadanayso kediska.

1. Waa maxay shaqada ugu weyn ee cod-bixiyeha ku jira moodeel-decoder-gaabiyaha?
   • Si aad u akhrido gelinta oo aad u rogto matalaad gudaha hodan ah
   • Si loo dhaliyo wax soo saarka hal calaamad markiiba
   • Si loo kala saaro natiijada musharraxiinta ixtimaalka
   • Si loo kaydiyo xogta tababarka tusaalaha
2. Muxuu cod-bixiyeyaashu u adeegsadaan is-fiirsashada (sababta) waji-xidhan?
   • Si loo qariyo gelinta codeeyaha
   • Si looga hortago boos kasta inuu arko calaamadaha mustaqbalka inta lagu jiro jiilka
   • Si loo cadaadiyo wax soo saarka calaamado yar
   • In si aan kala sooc lahayn loo tuuro calaamadaha si joogto ah
3. Farsamaynkee ayaa isku xira furaha iyo matalaada codeeyaha?
   • Ka saarida
   • Dufcaddii caadiga ahayd
   • Fiiro gaar ah
   • Codaynta booska
4. Midkee ka mid ah moodooyinkan waa encoder-decoder (isku xigxiga-ilaa-taxan) qaab dhismeedka?
   • GPT-3
   • BERT
   • Word2Vec
   • T5
5. Waa maxay sababta nashqadeeyaha-decoder-ku uu ugu habboon yahay tarjumaada dabiiciga ah?
   • Sababtoo ah wax-soo-saarka iyo soo-saarku waa taxane kala duwan, oo u baahan faham buuxa ka dibna jiil cusub
   • Sababtoo ah gelinta iyo soo saarista mar walba waa isku dherer
   • Sababtoo ah waligeed uma baahna inay soo saarto qoraal cusub
   • Sababtoo ah waxay ka fogaataa isticmaalka dareenka gebi ahaanba
6. In the original Transformer architecture, what attention does the encoder use over its input sequence?
   • Causal attention that hides all earlier source tokens
   • Bidirectional self-attention over the full source sequence
   • Cross-attention only to an image sequence
   • No attention mechanism
7. What training objective does BART use for pretraining?
   • It predicts only the next pixel in an image
   • It classifies each word into a fixed sentiment label
   • It corrupts text and trains a sequence-to-sequence denoising model to reconstruct the original
   • It performs speech recognition from raw audio
8. How does T5 frame many NLP tasks for one encoder-decoder model?
   • As image classification with a shared label set
   • As speech waveforms with no text tokens
   • As separate model architectures for each task
   • As text-to-text inputs and outputs with task descriptions in the input text