quiz bu lëkkale ak tegtal · Dëgër Niveau

xët yi ak quiz vLLM

Test PagedAttention KV blocs, kartograafi logique-à-physique, ak séddoo, boole ci wuute gi nekk ci digganté dencukaay cache ak dencukaay model bi yépp.

Yooni njiit yu jëm ciPagedattention Ak Vllm
Laaj 1 de 8

Lan la 'KV cache' di denc ci defar mbind?

Xoolaatal laaj bu nekk8 laaj

Sooy def quiz bi, tontu yi ak leeral yi dañuy feeñ.

  1. Lan la 'KV cache' di denc ci defar mbind?
    • Done yiñ tàggat ci model bi
    • Njàngalem jëfandikukat bi
    • Caabi ak vecteur valeur ngir jetons yu njëkk yi suko defee jetons yu bees yi mëna leen toppatoo
    • ab kopi buñ komprime ci diisaayu model bi
  2. Ban xalaatu sistem jëfandikoo moo inspiré PagedAttention?
    • Jumtukaayi defragmentaasioŋ disk
    • Paging mémoire virtuel
    • Kompresioŋ fichier
    • CPU bànxaasu
  3. Lan mooy jafe-jafe bi PageAttention mëna saafara ci wàllu cache KV?
    • Koneksioŋ internet yu yeex
    • Ñàkum done ci wàllu tàggat yaram
    • Kaaraange baatu-jàll bu néew doole
    • Yaxxum mémoire biy bawoo ci dencu ay blok yu yaatu yu jegewaale ngir laaj bu nekk
  4. naka lay def ba PagedAttention may génne yu bari ñu séddoo mémoire ngir benn laaj buñ bokk?
    • Xeeti toppalante yu wuute dañuy wane benn xët physique jaaraleko ci kopie-ci-bind
    • Dafay efaase laaj bi ginaaw biñu ko njëkkee jëfandikoo
    • Dafay denc laaj bi ci disk bi
    • Dafay tàggataat model bi ci génn gu nekk
  5. Fan lañu njëkka defar vLLM ak PagedAttention?
    • Laboratoire akademik ci ati 1990
    • UC Berkeley (ñu siiwal ko ci 2023)
    • Ekipu fësal jaaykatu aparey
    • Benn forum ci net bi te kenn xamul turam
  6. Lan mooy njariñ li gëna mag ci vLLM ci nit kuy liggéey LLM?
    • Modèle bu gëna mag ci wàllu référence
    • Etiketu done otomatik
    • GPU bu amul fayda
    • Production bu gëna mag ak laaj yu gëna bari ci GPU bu nekk
  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