Quiz biy waaja tàggat ELECTRA

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

1. Ban liggéey la ELECTRA di jëfandikoo ci palaasu modelu làkk bu maske?
   • ñu wecci gis-gis token
   • Prediction frase bi ci topp
   • Xalaatal baat bu ci topp buy dellu ginaaw
   • Modelu làkku permutasioŋ
2. Lu tax ELECTRA gëna am njariñ ci xayma BERT?
   • Dafay jëfandikoo vocabulaire bu gëna ndaw
   • Dafay xayma siñaal jàng ci kaw jeton yépp, du ci kaw yiñ maske rek
   • Dafay tàggat ci 15% ci jeton yiñ maske
   • Dafay teggi mekaniismu bàyyi xel bi
3. Ban cër la reso generatër bu ndaw bi di def ci ELECTRA?
   • Dafay rang produit yu mujj yi ci kalite
   • Dafay kompresse model bi ginaaw biñu ko tàggatee
   • Dafay defar ay token yuñ mëna wecci ngir kiy xeex mën àtte
   • Dafay tekki mbind diggante làkk yi
4. Lan mooy dall generatër bi ginaaw bi ñu tàggatee ba noppi?
   • Dañu ko jagleel liggéey bi ci suuf
   • Dafa boole ak diskriminatër bi
   • Dafay nekk decodeur ngir jamono
   • Dañu koy sànni; diskriminatër bi kese lañuy jëfandikoo
5. Gëstukat yu bawoo ci ban mbootaay ñoo njëkka defar ELECTRA?
   • Google and Stanford
   • AI ak NYU
   • OpenAI ak Microsoft
   • Deepmind ak Oxford
6. Ban model ci ginaaw moo yokk xalaatu ELECTRA ci gis-token buñ wecci?
   • GPT-2
   • Deberta-v3
   • T5
   • Word2Vec
7. What happens if ELECTRA’s generator proposes the original token at a position?
   • It is always labeled replaced because the generator proposed it
   • The position is removed from the input
   • That token is treated as original by the discriminator’s replaced-token label
   • The whole sequence is discarded
8. When comparing ELECTRA with another pretraining method, what makes a compute claim meaningful?
   • Compare model names alone
   • Assume every ELECTRA run uses one fixed compute ratio
   • Ignore generator training cost
   • Specify model size, data, training budget, and evaluation setting