LLM-sida-a-Xaakim Kedis ah

Jawaabaha iyo sharraxaadaha waxay soo baxayaan markaad qaadanayso kediska.

1. Waa maxay 'LLM-as-a-judge' loola jeedaa?
   • Isticmaalka hal nooc oo luqadeed si aad u dhaliso ama barbar dhigto wax soo saarka nooc kale
   • Isticmaalka qaabka luqadda si aad u qorto xukunnada sharciga ah
   • Qaabka kaliya ee ka jawaaba haa/maya su'aalaha
   • Qaabka qiimaynaya natiijooyinka raadinta taariikhda
2. Waa maxay faraqa udhaxeeya garnaqsi fiiqan iyo laba-geesood leh?
   • Pointwise waa dhakhso badan; labada dhinacba way ka gaabiyaan laakiin waa isku mid haddii kale
   • Si toos ah ayaa u qiimeeya hal jawaab oo ka soo horjeeda qoraal; si laba-dhammaan ah u doorta laba ka wanaagsan
   • Pointwise kaliya waxay u shaqeysaa koodka; laba-laaban yahay oo keliya tiraab
   • Ma jiro farqi dhab ah
3. Midkee ayaa ah eexda si wanaagsan loo diiwaangeliyay ee garsoorayaasha LLM?
   • Doorbidaaya jawaabaha ku qoran dhammaan xarfaha waaweyn
   • Had iyo jeer door bidaya kelmadda koowaad alifbeetada
   • U roonaanta jawaabaha dhaadheer iyadoon loo eegin tayada
   • Diidmada darajada su'aalaha ku saabsan taariikhda
4. Sidee bay qiimeeyayaashu caadi ahaan uga hortagaan eexda marka la barbardhigo labada dhinac?
   • Adigoo tusaya hal jawaab markiiba
   • Adigoo mar walba ka hor marinaya jawaabta dheer
   • Adiga oo tirtiraya goobta xaq-siinta
   • Adigoo ku socodsiiya isbarbardhig kasta laba jeer iyadoo amarka la beddelay
5. What evaluation procedure did G-Eval use for generated text?
   • Chain-of-thought steps followed by a form-filling evaluation against criteria
   • A speaker verification threshold
   • A search index with BM25 only
   • An image classifier trained on screenshots
6. Sidee kooxi ku xaqiijin kartaa in xaakimka LLM uu la socdo dookhyada aadanaha?
   • Adigoo waligaa dib u hubin ka dib dejinta
   • Iyada oo la barbardhigayo set dahab ah oo bini'aadmigu ku suntan yahay
   • Adigoo ka dhigaya garsooraha laftiisa darajo
   • Adigoo isticmaalaya kaliya maalmaha fasaxa ah
7. What is self-preference bias in LLM-based evaluation?
   • A judge always selects the shortest answer
   • A judge cannot compare two outputs
   • A judge favors outputs generated by its own model or model family
   • A judge evaluates only audio signals
8. What does G-Eval use to make its discrete score more fine-grained in the reported method?
   • The position of the answer in a pairwise comparison
   • The judge model’s parameter count only
   • A count of citations in the answer
   • Probability weights over the candidate score tokens