Kediska Modelinka Mawduuca-dheer

Jawaabaha iyo sharraxaadaha waxay soo baxayaan markaad qaadanayso kediska.

1. Muxuu yahay 'daaqadda macnaha guud' ee moodeelka?
   • Tirada ugu badan ee calaamaduhu waxay ku geli kartaa hal baas
   • Cabbirka shaashadda ee interface chat
   • Tirada isticmaalayaasha la oggolaaday hal mar
   • Ilaa intee la tababaray qaabka
2. Waa maxay sababta is-fiirinta caadiga ah ay qaali ugu noqoto wax-soo-saarka dheer?
   • Waxay u kortaa si toos ah mar walbana waa raqiis
   • Xisaabinteeda iyo xusuusteeda waxay u koraan afar-geesood iyagoo leh dherer isku xiga
   • Ma akhrin karo wax ka badan hal calaamad
   • Waxay kaliya ku shaqeysaa sawirada
3. Waa maxay dhibaatada 'dhexda ka luntay'?
   • Moodooyinka waxay illoobaan gabi ahaanba tababarkooda
   • Qaababku waxay akhrin karaan oo keliya badhtamaha dukumeentiga
   • Qaababku waxay dib u xasuustaan macluumaadka bilowga iyo dhammaadka wax gelinta dheer si ka wanaagsan xaqiiqooyinka dhexda lagu aasay
   • Qaababka ayaa dhexda u tarjuma luqad kale
4. Maxay FlashAttention ugu horrayn u horumarisaa?
   • Waxay ka dhigaysa soo-saarka moodalka qoraal dheer
   • Waxay si toos ah ugu daraa xog tababar oo badan
   • Waxay meesha ka saaraysaa baahida loo qabo dareen kasta
   • Waxay ku yaraynaysaa murugada xusuusta iyadoo la adeegsanayo IO-ware, xisaabinta foornada
5. Door noocee ah ayaa ku-xidhka booska rotary-ga (RoPE) ka ciyaara moodooyinka muddada-dheer?
   • Waxay calaamadeeyaan boosaska calaamadaynta iyo, isdhexgalka, waxay caawiyaan guud ahaan taxanaha dheer
   • Waxay ka tirtiraan calaamadihii gelinta
   • Waxay u turjumaan qoraalka luqadaha dhexdooda
   • Waxay kaydiyaan miisaanka modelka
6. What does prompt-cache research reuse across compatible requests?
   • The final answer regardless of the new query
   • Computed attention state for reusable prompt segments
   • A fixed set of token IDs without model computation
   • Xogta tababarka moodeelka
7. What does the RULER benchmark add beyond locating one isolated “needle” fact?
   • Kala soocida sawirka kaliya
   • Audio transcription from waveforms
   • Tasks such as multi-hop tracing and aggregation over long contexts
   • GPU memory allocation benchmarks
8. Why test several task types when measuring a model’s effective context length?
   • Every task uses the same computation
   • The advertised window guarantees all tasks work
   • Task type never affects context use
   • A model may handle simple retrieval yet fail to combine or track information across a long input