Kediska Fiiro-dhaafka

Jawaabaha iyo sharraxaadaha waxay soo baxayaan markaad qaadanayso kediska.

1. Waa maxay farqiga ugu muhiimsan ee u dhexeeya is-fiirinta iyo is-fiirinta?
   • Fiiro gaar ah, weydiimaha waxay ka yimaadaan hal taxane iyo furayaal/qiimo kale
   • Fiiro gaar ah ma isticmaalo softmax
   • Feejignaanta nafta waxay ku shaqeysaa oo keliya sawirrada
   • Feejignaanta isdhaafka ah laguma tababari karo asal tartiib tartiib ah
2. Beddelka-dekodeeriyaha, halkee ayay ka yimaadaan weydiimaha feejignaanta?
   • Soo saarida codeeyaha
   • Xaaladda hadda jirta ee codeeyaha
   • Gelida cayriin kaliya
   • Matrix random gaar ah
3. Waa maxay sababta labada tixood ee isku-tallaabta ah ay u yeelan karaan dherer ama qaabab kala duwan?
   • Sababtoo ah dareenku wuxuu u baahan yahay wax-soo-galin dherer siman
   • Sababtoo ah waxay iska indha tirtay mid ka mid ah taxanaha gebi ahaanba
   • Sababtoo ah su'aalaha iyo furayaashu waxay ka yimaadaan ilo kala duwan, oo madax-bannaan
   • Sababtoo ah waxay marka hore wax walba ku dhejisaa cabbir go'an
4. Sidee buu Stable Diffusion u adeegsadaa feejignaanta?
   • Si loo cadaadiyo sawirada faylal yaryar
   • Si loo turjumo qoraalka u dhexeeya luqadaha
   • Si looga saaro buuqa codka
   • Si loo sharxo gobollada sawirka la sameeyay ee isla markiiba qoraalka
5. Waa maxay hawlgalka xisaabeed ee isdhaafsiga la wadaago is-fiirsashada?
   • Feejignaanta dhibcaha-soosaarka la miisaamay oo leh softmax
   • Isbadal ka badan pixels
   • Cusboonaysiinta gobolka ee soo noqnoqda
   • Miisaska raadinta summada adag
6. What gate does Flamingo use to control information flow through its cross-attention layers?
   • A hard threshold initialized to one
   • A tanh gate initialized to zero
   • A fixed residual weight of 0.5
   • A learned vocabulary lookup table
7. What input sequence structure does Flamingo’s paper say its visual-language model can handle?
   • Only one image followed by a fixed-length label
   • Only text tokens with no visual inputs
   • Visual and textual data interleaved in arbitrary sequences
   • Only video frames with no text
8. What is one purpose of cross-attention in Flamingo’s visual-language architecture?
   • It replaces all text tokens with pixels
   • It prevents any information from passing between modalities
   • It performs only image resizing
   • It lets language-model layers incorporate visual features from image/video inputs