Kedis ku xidhan hagaha · Adag Heerka

Paged fiiro gaar ah iyo vLLM Kedis

Tijaabi Boggad Feejignaanta KV blocks, khariidad-jiseed macquul ah, iyo wadaagis, oo ay ku jirto farqiga u dhexeeya kaydinta xusuusta kaydinta iyo wadarta kaydinta moodalka.

Wadooyinka haga ee la xidhiidhaDareenka Bogga iyo Vllm
Su'aal 1 ee 8

Muxuu kaydiyaa 'KV cache' inta lagu jiro qoraalka qoraalka?

Horudhac su'aal kasta8 su'aalo

Jawaabaha iyo sharraxaadaha waxay soo baxayaan markaad qaadanayso kediska.

  1. Muxuu kaydiyaa 'KV cache' inta lagu jiro qoraalka qoraalka?
    • Xogta tababarka modelka
    • Aqoonsiga gelitaanka isticmaalaha
    • Furaha iyo qiimaha calaamadihii hore si calaamooyinka cusubi ay uga qaybgalaan iyaga
    • Nuqul la cufan oo miisaannada moodeelka ah
  2. Waa maxay fikradda nidaamka hawlgalka dhiirigeliyay PagedAttention?
    • Qalabka defragmentation Disk
    • Xusuus-qorka Virtual
    • Cadaadiska faylka
    • Saadaasha laanta CPU
  3. Dhibaato noocee ah ayaa ku jirta qoondaynta kaydka KV ee dhaqameed ee PagedAttention xalliyo?
    • Xidhiidhada internetka oo gaabis ah
    • La'aanta xogta tababarka
    • Nabadgelyada erayga sirta ah oo daciif ah
    • Qashinka xusuusta ee kaydinta baloogyada waaweyn ee isku xidha codsi kasta
  4. Sidee PagedAttention u ogolaataa soosaarayaal badan inay wadaagaan xusuusta degdeg guud?
    • Tixanayaal kala duwani waxay tilmaamayaan isla bogag muuqaal ah iyada oo loo marayo nuqul-qoritaan
    • Waxay tirtirtaa degdegga ka dib isticmaalka ugu horreeya
    • Waxay ku kaydisaa degdega diskka
    • Waxay dib u habaynaysaa qaabka wax soo saar kasta
  5. Halkee laga sameeyay vLLM iyo PagedAttention markii hore?
    • Shaybaadhka akadeemiyadeed ee 1990-yadii
    • UC Berkeley (la daabacay 2023)
    • Kooxda suuq geynta iibiyaha qalabka
    • Gole online ah oo qarsoodi ah
  6. Waa maxay faa'iidada ugu weyn ee vLLM ee qof u adeegaya LLM?
    • Saxnaanta moodeelka ka weyn ee bartilmaameedyada
    • Calaamadaynta xogta tooska ah
    • Qalabka GPU ee bilaashka ah
    • Wax-soo-saarka sare iyo codsiyo badan oo isla socda halkii GPU
  7. A sequence’s logical KV block 3 resides in a different physical location. What tells PagedAttention where to read it?
    • The sequence’s block table mapping logical blocks to physical blocks
    • The model’s class-label list
    • The learning-rate scheduler
    • A requirement that every block follow block 2 contiguously
  8. A PagedAttention cache ends with a half-filled block. Which statement about memory waste is accurate?
    • Paging guarantees every allocated byte contains a token
    • The entire model must be copied for each unused slot
    • Unused slots can remain in that final block despite reduced allocation fragmentation
    • The block table cannot represent a partial sequence