Bixinta Feejignaanta iyo Kediska Goynta Madaxa

Jawaabaha iyo sharraxaadaha waxay soo baxayaan markaad qaadanayso kediska.

1. Waa maxay hadafka fiiro gaar ahaaneed?
   • Si loo raadiyo socodka xogta isugaynta ah ee ka imanaya calaamadihii gelinta ilaa wax soo saarka lakabyada
   • Si aad lakabyo badan ugu darto Transformer
   • Si loo kordhiyo heerka waxbarashada
   • Si dareenka loogu beddelo qalqal
2. Waa maxay sababta khariidadda dareenka lakabka ahi ay inta badan ugu filnayn sharraxaadda?
   • Feejignaanta marna lama xisaabiyo
   • Xogtu waxay isku dhafan tahay lakabyo iyo madaxyo badan, sidaa darteed hal khariidad ayaa seegtay dariiqa buuxa
   • Maabku had iyo jeer waa isku mid
   • Transformers waxay leeyihiin hal madax oo keliya
3. Maxay fiiro gaar ah u soo kordhinaysaa shax kasta oo feejignaan ah si loogu xisaabtamo xidhiidhada hadhsan?
   • Eex eber ah
   • Eray sanqadh random
   • Qayb aqoonsi ka hor inta aan caadi laga dhigin
   • softmax dheeraad ah
4. Maxay cilmi-baadhis ku samaysay feejignaan madax-badan ka muujisay madax-furnaanshiyaha?
   • Madax kasta waa lama huraan
   • Madax lama saari karo
   • Goynta had iyo jeer waxay labanlaabtaa saxnaanta
   • Madax badan ayaa lagu gooyn karaa iyada oo wax yar oo sax ah la waayo
5. Sidee muhiimada madaxu caadi ahaan loogu qiyaasaa manjooyinkeeda?
   • Iyadoo loo marayo dareenka ku salaysan gradient ee luminta maaskaro madaxa
   • Markay tahay meesheeda alifbeetada
   • Adiga oo tirinaya halbeegyadiisa kaliya
   • Waqtiga maalinta
6. What different jobs do attention rollout and head pruning each do?
   • Waa farsamooyin isku mid ah
   • Soo-bandhigiddu waxay caawisaa tarjumaada halka manjooyinku ay ka faa'iidaysan karaan ka-noqoshada si loo cadaadiyo moodooyinka
   • Labaduba waxay ku daraan oo kaliya halbeegyada
   • Midkoodna ma taabanayo dareenka
7. In an attention-rollout calculation, why are residual paths included with layer attention?
   • To make every head have the same weight
   • To remove the value vectors from the Transformer
   • To approximate information that can pass along the identity path when layer transitions are composed
   • To prevent the attention matrix from being normalized
8. How should an attention-rollout score be interpreted?
   • As proof that a token caused the output
   • As the exact value vector used at every layer
   • As a measure of how many heads the model contains
   • As a post-hoc approximation of relative token influence, not a causal explanation