Kediska Qiimaynta Tilmaamaha Baadhista Tooska ah iyo Barafka

Jawaabaha iyo sharraxaadaha waxay soo baxayaan markaad qaadanayso kediska.

1. Baaritaanka tooska ah ee caadiga ah, cabbirkee ayaa la cusboonaysiiyay inta lagu jiro tababarka?
   • Kaliya kala soocida toosan ee dhawaan lagu daray ee sare
   • Laf-dhabarta oo dhan iyo lakabka toosan
   • Kaliya laf dhabarta, ka tagista kala soocida mid go'an
   • Lakabka ku dhejinta iyo xannibaadda dareenka ugu dambeeya
2. What does high linear-probe accuracy most directly show?
   • The backbone causally uses the feature for every prediction
   • The labels are decodable by a linear classifier from the frozen features under that dataset and protocol
   • The representation is robust to every distribution shift
   • The probe has identified a human-interpretable mechanism
3. Marka la barbar dhigo hagaajinta buuxda, baaritaanka tooska ah ayaa ugu horayn soo jiidasho leh sababtoo ah waa:
   • Hawl kasta oo aad u sax ah
   • La dammaanad qaaday in si ka wanaagsan wax loo qaybiyo
   • Way ka jaban tahay xisaabinta iyo xusuusta waxayna siinaysaa akhrin nadiif ah oo tayo leh
   • Habka kaliya ee loo isticmaalo qaab horay loo tababaray
4. Helitaanka caamka ah ee caadiga ah marka la baarayo transformers waa:
   • Lakabka ugu dambeeya wuxuu had iyo jeer bixiyaa saxnaanta baaritaanka ugu fiican
   • Baaritaanku waxa uu ka shaqeeyaa shabakadaha isdhaafsiga
   • Doorashada lakabka weligood ma saameeyaan natiijooyinka baaritaanka
   • Lakabyada dhexe ayaa inta badan ka sarreeya lakabka ugu dambeeya ee wareejinta
5. Marka laga soo tago bar-tilmaameedka, baaritaannada toosan ayaa si sii kordheysa loogu isticmaalay fasiraadda:
   • Ogow in fikradda runta ah ay ku qoran tahay dalalka qarsoon
   • Nadiifi moodooyinka loogu talagalay geynta mobilada
   • Beddel wanaajinta inta lagu jiro tababarka ka hor
   • U samee xogta tababarka cusub si toos ah
6. Khatarta habka muhiimka ah marka la turjumayo saxsanaanta baadhista sare waa:
   • Laf-dhabarta ayaa si qarsoodi ah u ilaalisa tababarka
   • Baadhitaanka ayaa laga yaabaa in laga faa'iidaysto isku xidhka beenta ah ee aan ahayn fikradda runta ah
   • Baarayaasha toosan ma dhaafi karaan
   • Khasaare isdhaafsi-entropy lama isticmaali karo
7. To compare two backbone layers with linear probes, what should remain consistent?
   • The feature vectors from each layer
   • The layer being evaluated
   • The task labels, data split and probe-training protocol
   • The activation values themselves
8. Why might feature normalization and learning-rate sweeps be used in a probe evaluation?
   • To update the frozen backbone
   • To add hidden nonlinear layers to the probe
   • To guarantee the concept is causally used
   • To reduce optimization and scale effects that could distort the representation comparison