quiz bu lëkkale ak tegtal · Diggante Niveau

Quiz distilaasioŋ xam-xam

Jëfandikool séddale jàngalekat yi, mébet yu woyof yi, ak jàngat ndongo yi ci distilaasioŋ, boole ci lu tax toppandoo mënul garanti njubte ci niveau jàngalekat bi.

Yooni njiit yu jëm ciDistilaasioŋ Xam-xam
Laaj 1 de 8

Ci distilaasioŋ xam-xam, lan lañu tàggat ndongo yi ñu def?

Xoolaatal laaj bu nekk8 laaj

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

  1. Ci distilaasioŋ xam-xam, lan lañu tàggat ndongo yi ñu def?
    • Taggat ndongo bu ndaw mu toppandoo ab jàngalekat bu mag
    • Ngir defar xeetu model bu gëna rëy te gëna jub
    • Ngir etiketu done ci saasi
    • ngir enkripte diisaayu benn model
  2. Luy 'xam-xam bu lëndëm' bu jàngalekat bi?
    • Jeffin ju ñaaw ju nëbbu ci xeetu mbir mi
    • Wàll wiñ enkripte ci fichier model bi
    • Leeral yi ci probabilite relatif yi jàngalekat bi jox klaas yu jaarul yoon
    • Done yi jàngalekat bi musul tàggat ci
  3. Ban cër la tàngoor (T) di def ci distilaasioŋ?
    • Dafay saytu gaawaayu tàggat model bi kese
    • Dafay nooy softmax suko defee nuru yu ndaw yi am ci diggante klaas yi nekk luñu mëna jàng
    • Dafay tëral limu diisaay yi ci ndongo mi
    • Dafay natt tàngoor wi GPU bi di génne
  4. Ban ñaari mbir ñoo boole perte distillation bu yàgg bi?
    • Ñaari xeeti neexal yu wuute
    • Jumtukaay kaare moyen ak ñàkka am charnière
    • Daan KL ci wàllu politiku royuwaay ak fonction valeur
    • Ab term KL-divergence bu méngoo ak génnug jàngalekat bi boole ci entropie croix ci etiketu dëgg yi
  5. Lan mooy misaalu xeetu làkk buñ distiye?
    • Resnet-50
    • Word2Vec
    • distilla
    • AlexNet
  6. Lan moo wuutale distillation bu sukkandiko ci man-man yi ak anam wiñ ko baaxe ci génne gi?
    • Dafay méngoo ak diggu gis-gis yu nëbbu yi, te baña yam ci génne yu mujj yi
    • Dafay baña bàyyi xel ci jàngalekat bi
    • Dafay jëfandikoo etiket yu dëgër kese
    • Dafay tàggat jàngalekat bi, du ndongo li
  7. A teacher assigns similar small probabilities to wolf and fox when predicting dog. What can soft-target distillation transfer beyond the top label?
    • A guarantee that either alternative is the correct label
    • The teacher’s complete original training corpus
    • The teacher’s exact number of hidden layers
    • Relative relationships among the teacher’s alternative class scores
  8. A distilled student is smaller than its teacher. Which evaluation is needed before deployment?
    • Assume its accuracy must equal the teacher’s
    • Measure its task accuracy and runtime on the intended system
    • Assume fewer parameters guarantee lower latency on every device
    • Only compare the models’ names