Meta-Jàngale

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

1. Lan mooy meta-learning, wala 'jàng ngir jàng', di njëkka am?
   • Gaaw ànd ak liggéey yu bees ci misaal yu néew
   • Memoriser benn ensemble done bu mag
   • Rax ci dolli soxla tàggat yaram
   • kompresse model ci dayo zero
2. Bu ñuy meta-training, naka lañuy faral di defaree done yi?
   • Benn liggéey bu amul benn xaaj
   • Liggéey yu bari, bu nekk ak setu ndimbal ak setu laaj
   • Benn nataal baamtu
   • Ni bruit buñu etiketewul rek
3. Luy MAML jàng?
   • Classificateur fixe bu dul soppeeku
   • Done yu negatif
   • Benn xeetu tàmbali buñ mëna méngale ak liggéey yu bees ci jéego yu néew
   • Kodeku kompresioŋ
4. naka lañuy xaajalee misaal yu bees yi ci reso prototypique yi?
   • Ni ñuy defaree nataal yu bees
   • Ni ñuy jàppee valeur pixel yi
   • ci xalaat yu bari
   • Soo leen méngale ak prototype yiñ jàngee ci biir barab buñ samp
5. Lan mooy wareefu bouclage bi ci biir MAML?
   • Ngir méngale model bi ak benn liggéey ak yenn jéego yu am degrade
   • ngir jaxase done yi
   • Ngir efaase ay parametre
   • Def taxawaayu jàng ci nul
6. How does GPT-3 perform the few-shot demonstrations in the cited paper?
   • By updating all model weights after every example
   • By including examples in the prompt without gradient updates or fine-tuning
   • By storing one prototype per class
   • By running a MAML inner loop during inference
7. In Reptile, how is the meta-initialization updated after adapting on a sampled task?
   • Replace the model with a nearest-neighbor table
   • Average class labels without updating parameters
   • Move the initialization toward the task-adapted parameters
   • Train a separate model for every example
8. What computational shortcut distinguishes Reptile from the second-order MAML formulation?
   • It requires no gradient steps
   • It stores only class prototypes
   • It trains without tasks or task-specific data
   • It can use a first-order update without computing second-order derivatives