quiz bu lëkkale ak tegtal · Diggante Niveau

GPTQ ak AWQ Quiz buy xayma limu ginaaw tàggat yaram

Saytu wuute gi nekk ci digganté GPTQ ak AWQ, done etalonnage, kompensaasioŋu courbure, chaine yiy xam aktivasioŋ ak mémoire bu néew bit.

Yooni njiit yu jëm ciGptq ak Awq ginaaw tàggat
Laaj 1 de 8

Luy 'kantisation ginaaw tàggat yaram'?

Xoolaatal laaj bu nekk8 laaj

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

  1. Luy 'kantisation ginaaw tàggat yaram'?
    • Komprime diisaayu model buñu jota tàggat ngir wàññi njub te doo tàggataat bu baax
    • Taggat model bi lëmm ci anam wu woyof
    • Yokk yeneen done tàggat yaram ginaaw biñu ko dugalee
    • Yokk limu parametru model bi
  2. Ban leeral la GPTQ di jëfandikoo ngir dindi njuumti yi ci wàllu wërsëg suñuy xayma?
    • Limu baatu model bi
    • Ñaareelu rang (Hessian/courbure) leeral ci benn done buñ etalonnage
    • Taarixu laajtu jëfandikukat bi
    • Bruit bu bari
  3. Lan mooy gis-gis bi gëna am solo biy dawal AWQ (Kantasioŋ diisaay biy xam-xam)?
    • Diisaay yépp a am solo
    • Activation yi deñu leen wara quantifié waaye poids yi duñu laal
    • Ndaw yu ndaw ci chaine de poids yi dañu am solo lool te dañu leen wara aar
    • Kantitasioŋ war na am ci diiru tàggat yaram kese
  4. Lu tollu ci ñaata mémoire la quantisation 4-bit di sakkanal ak poids 16-bit yi?
    • Amul xaalis buñ denc
    • Lu tollu ci ñaari yoon mémoire bi
    • Genn-wàll ci mémoire bi
    • Lu tollu ci benn ci ñatti pàcci mémoire bi (4x wàññeeku)
  5. Lan moo waral GPTQ ak AWQ ñoom ñaar ñuy xayma diisaay yi waaye ñuy tëye aktivaasioŋ yi ci gëna dëggu?
    • Poid yi ñoo ëpp doole ci mémoire bi, te kantite biy aktive bu metti dafay gëna gaañ jaar-jaar bi
    • Aktiwisaasioŋ amul ci transformatër yi
    • Aktiwisaasioŋ yi 1-bit lañu xaat
    • Yoon daf ko sàkku
  6. Ban jumtukaay lañuy faral di jëfandikoo ngir doxal ay model yuñ xayma ci barab bi ci hardware konsomatër bi?
    • Word
    • llama.cpp
    • Prograamu këyitu xayma
    • Navigateur web kese
  7. How does GPTQ use a calibration dataset during layer-wise weight quantization?
    • It adds new training labels to the model
    • It selects a new tokenizer vocabulary
    • It estimates second-order information to compensate for rounding errors
    • It measures only GPU temperature
  8. How does AWQ identify which weight channels deserve protection?
    • It computes a full Hessian inverse
    • It adds random noise to all weights
    • It selects channels only by their index
    • It uses activation magnitudes to identify channels important to preserve