Jàng mu wéy ak quiz fàtte musiba

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

1. Luy 'fatte bu metti'?
   • Soo jàngee liggéey bu bees, loolu dafay wàññi bu baax sa kàttan ci liggéey yi nga jàngoon bu njëkk
   • Su benn model jeexee memory ba noppi mu tass
   • Suñu efaasee done tàggat yaram ci anam wu jaarul yoon
   • Su benn model bañee wax luy am
2. Li ñuy woowe kompromis bi nekk ci xol njàng mu wéy mooy:
   • kompromis bias-variance
   • Jafe-jafe dayo ak plastisite
   • Kompromis gaawaay-jub
   • kompromis diggante wër ak jëfandikoo
3. Luy boole poid elastik (EWC)?
   • efaase ay neuron yu amul solo
   • Yokk ñaari yoon limu jàng ci liggéey yu bees
   • Dafay yar coppite yi am ci poid yi am solo ci liggéey yi ñu jàngoon bu njëkk
   • Dafay denc bépp misaalu tàggat yaram bu njëkk ba fàww
4. naka la pexe yu lalu ci replay di xeexe fàtte?
   • Sooy congelate reso bi yépp ba fàww
   • Ni ñuy jëfandikoo tolluwaayu jàng bu gëna mag
   • Soo dindi fonction perte
   • Ni ñuy boolee misaal yu yàgg yiñ denc wala yuñ defar ci tàggat bu bees
5. How does Gradient Episodic Memory (GEM) constrain a gradient update while learning a new task?
   • It projects the gradient so its first-order approximation does not increase loss on stored prior-task examples
   • It replaces all prior-task data with new examples
   • It freezes every parameter for all future tasks
   • It increases weights that were unimportant to earlier tasks
6. Lan moo waral ñu fàtte liggéey bu bees bi nga def ci liggéey bu bees?
   • GPU bi dafay tàng lool ba noppi yàq mémoire bi
   • Wàccig gradient dafay toxal diisaay yiñ bokk ci liggéey bu bees bi te amul benn tere ngir denc yu yàgg yi
   • Liggéey yu bees dañuy faral di am done yu bari
   • Modèle bi dafay efaase poid yu yàgg yi
7. How does iCaRL classify a test example after learning multiple classes?
   • By retrieving the single closest word in a text corpus
   • By selecting the class with the largest Fisher information
   • By comparing its feature vector with per-class means computed from stored exemplars
   • By replaying every training image at inference
8. Why is evaluating continual learning more than measuring final accuracy on the newest task?
   • New-task accuracy is the only relevant measure
   • Old tasks are removed from the evaluation
   • The model cannot be tested on multiple tasks
   • A method must also measure retention on prior tasks and its ability to learn new ones