Decoding mala awaal ah oo leh EAGLE Quiz

Jawaabaha iyo sharraxaadaha waxay soo baxayaan markaad qaadanayso kediska.

1. Waa maxay fikradda udub-dhexaadka ah ee ka dambeysa dejinta malo-awaalka?
   • Qore raqiis ah ayaa soo jeedinaya calaamado badan oo qaabka weyni ku caddeeyo hal baas oo barbar socda
   • Cadaadiska miisaanka tusaalaha ilaa nuska saxda ah
   • Ka boodista lakabyada shabakada si aan kala sooc lahayn
   • Horay u xafidi jumlad kasta oo suurtogal ah
2. Sidee ayuu EAGLE kaga duwan yahay hababkii hore ee codaynta?
   • Waxay meesha ka saaraysaa tallaabada xaqiijinta gebi ahaanba
   • Waxay u qaabaynaysaa qaabka qarsoon ee booska halkii ay toos uga ahaan lahayd heerka calaamadda
   • Waxa ay u isticmaashaa qaab gebi ahaanba gooni ah, oo ka weyn qaab-sameeyaha
   • Waxay u shaqeysaa oo kaliya jiilka sawirka
3. Why is speculative decoding considered lossless when implemented with exact verification?
   • Every draft token is accepted unconditionally
   • The drafter always predicts the target’s next token
   • The acceptance and correction procedure preserves the target model’s output distribution
   • The target model is never evaluated
4. Dhibaato noocee ah ayaa ka jirta qaabaynta sifada EAGLE oo dib u soo celinta calaamadii hore ee ku-xidhiddisa ay caawiso?
   • Kala qaybsanaanta xusuusta ee GPU-ga
   • Tokenizer-ku isma barbar dhigin luqadaha
   • Kululaynta qalabka
   • Hubanti la'aanta / madmadowga saadaasha muuqaalka soo socda
5. Muxuu EAGLE-2 ku soo kordhiyay EAGLE-kii asalka ahaa?
   • Geed qabyo ah oo firfircoon oo la qabsanaya inta musharax ee soo jeedinta
   • Qabyo hal calaamad go'an
   • Ka saarida madaxa LM
   • Tababarka ku saabsan laba jeer xogta kaliya
6. Waa maxay sababta cidhifyada autoregressive ay jiilka LLM ee cad uga dhigayaan mid gaabis ah?
   • Qaabka ayaa dib isu tababara ka dib calaamad kasta
   • Calaamad kasta oo cusub waxay u baahan tahay gudbin hore oo buuxda oo balaayiin ah oo cabbiraad ah ka hor intaysan kan xiga bilaabin
   • Calaamadaha waxaa lagu kaydiyaa saxanka inta u dhaxaysa tallaabooyinka
   • GPU-gu waa inuu dib u kiciyaa calaamo kasta
7. After EAGLE predicts a hidden-state feature, which component turns that feature into candidate token probabilities?
   • The tokenizer’s vocabulary file alone
   • The optimizer’s momentum buffer
   • The target model’s language-model head
   • The KV-cache allocator
8. What change does EAGLE-3 make to the original feature-drafting constraint?
   • It replaces the target model with a tokenizer
   • Waxay ka joojisaa xaqiijinta bartilmaameedka
   • Waxay u baahan tahay qoraa bini'aadam oo gooni ah
   • Waxay ka saartaa xayiraadda saadaasha astaamaha si loo hagaajiyo ballaarinta