quiz bu lëkkale ak tegtal · Dëgër Niveau
Quiz jàng buy wuutale
Jàngale ñaari mbir yu baax yuy wuutale, daanu buñ samp, ak nuru ci tàngoor, boole ci ni raŋ MoCo di wuutee ak cache inference yi.
Laaj 1 de 8
Lan mooy mébetu jàngat bu wuute?
Xoolaatal laaj bu nekk8 laaj
Sooy def quiz bi, tontu yi ak leeral yi dañuy feeñ.
- Lan mooy mébetu jàngat bu wuute?
- Booleel ñaari mbir yu nuru (positif) nga puus ñaari mbir yu wuute (negatif) ñu tàqaloo ci biir bayaal bi
- Xalaatal valeur pixel yi ci benn nataal
- Xaajale bépp dugal ci benn ci etiket yiñ tëral
- Wàññil limu paramet model yi
- Ci anamu nataal buñuy saytu sa bopp bu melni SimCLR, naka lañuy sos peer bu baax?
- Ñaari nataal yu wuute ci benn klaas buñ etikete
- Benn nataal ak benn nataal buñu tann ci anam wu mucc ci dataset bi
- Ñaari yokkute yu bari (lu melni, mbay, jitter melo) ci benn nataal
- Benn nataal buñ boole ak masku segmentaasioŋ
- Ban fonction perte moo gëna méngoo ak jàng bu wuute?
- Jumtukaay kaare moyen
- Xibaar
- Entropie croisée ci kaw etiketu klaas yi
- perte de charnière pour SVMs
- Lan moo waral pexe yu melni MoCo di jëfandikoo bànku mémoire wala rang?
- Ngir denc etiketu ajustement buñ saytu
- Ngir denc waxtaan yu mujj yi ngir gëna gaaw
- ngir kompresse model bi ngir aparey mobile
- Ngir joxe misaal yu bari yu baaxul te soxlawoo dayo lote bu rëy
- Ban jafe-jafe la BYOL ak SimSiam di moytu te duñu jëfandikoo ay negatif yu leer?
- Overfitting ci etiketu tàggat yaram
- Degrade yuy kalaate ci optimisatër bi
- Representation dafay daanu, fu embedding yépp di nuru
- Tolluwaayu jàng buy jeex
- naka la CLIP di yokke xalaat bu wuute bi?
- Dafay méngale nataal yi ak seeni mbind ci barab buñ bokk
- Dafay wuutale ñaari yokkute ci benn clip audio
- Dina dindi bépp soxla ci reso neuronal
- Dafay jëfandikoo jàng buy dooleel ludul nuru
- A contrastive image encoder maps every photo to the same vector. Which failure does this illustrate?
- A useful separation of image categories
- An increase in the number of negative samples
- A perfect nearest-neighbor representation
- Representation collapse
- Why is MoCo’s training queue different from caching completed inference responses?
- It serves final chatbot responses without model execution
- It supplies encoded comparison examples for a contrastive training objective
- It replaces every positive image with a label
- It removes the need to update any encoder