Ikibazo gihuza ibibazo · Biragoye Urwego

Ikibazo Cyimvange na Ntangarugero Ikibazo

Ongera usuzume inzira ya Mixtral, gukora impuguke, igiteranyo hamwe nibintu bikora, hamwe nibikorwa bifatika byo kwibuka no kohereza ibicuruzwa bidasanzwe.

Inzira zijyanye no kuyoboraImvange kandi Ntangarugero
Ikibazo 1 Bya 8

Muri Mixtral 8x7B, abahanga bangahe batunganya buri kimenyetso?

Reba buri kibazo8 ibibazo

Ibisubizo nibisobanuro bigaragara mugihe ufata ikibazo.

  1. Muri Mixtral 8x7B, abahanga bangahe batunganya buri kimenyetso?
    • Impuguke ebyiri
    • Abahanga bose uko ari umunani
    • Impuguke imwe
    • Umubare uteganijwe buri gihe
  2. Niki kintu kigena impuguke ikimenyetso cyoherejwe?
    • Ikimenyetso
    • Router (umuyoboro winjira)
    • Icyerekezo
    • Imbonerahamwe
  3. Nubwo Mixtral 8x7B ifite ibipimo bigera kuri 47B byose, hafi ni bangahe bakora kuri token?
    • Hafi miliyari 47
    • Hafi miliyari 7
    • Hafi miliyari 13
    • Hafi miliyari 70
  4. What must a Mixtral serving setup account for even when only two experts run per token?
    • Only the active parameters ever need to be stored anywhere
    • All experts must always fit in GPU memory with no offloading option
    • Unused experts can be deleted without changing the checkpoint
    • The full expert weights must be available to the system, with residency or offloading depending on deployment
  5. Ni uruhe ruhushya Mistral AI yarekuye Mixtral, bikongerera uburenganzira bwo kwakirwa?
    • Apache 2.0
    • Umutungo, ufunze-isoko gusa
    • GPL v3
    • Ibikorwa bihanga Ibidandazwa
  6. What training steps does Mistral report for the Mixtral 8x7B Instruct variant?
    • Ikuraho ibikenewe byamakuru ayo ari yo yose
    • Supervised fine-tuning and direct preference optimization (DPO)
    • Bikuraho gukenera GPU burundu
    • Iremeza ko moderi itigera ikora amakosa
  7. How many expert networks are available in each Mixtral 8x7B sparse mixture-of-experts layer?
    • Impuguke ebyiri
    • Seven experts
    • Eight experts
    • Twelve experts
  8. Which languages does Mistral list for Mixtral 8x7B?
    • Yes; all unused expert weights are deleted
    • Yes; it means the model is dense with 12.9B total parameters
    • No; it means every parameter is active for every token
    • English, French, Italian, German, and Spanish