Laajteb etalonnage probabilite

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

1. Benn model dafay kalibre bu baax sudee, ci biir waxtaan yi dafa am 80% wóolu, xew-xew bi am:
   • Lu tollu ci 80% ci jamono
   • 100% ci jamono ji
   • Lu tollu ci 20% ci jamono
   • Benn yoon
2. Lan la tàngoor di def ci génnug reso neuronal bi?
   • Coppite ci ban klaas lañu séentu
   • Séddale logits yi ak scalar buñu jàng balaa softmax, di yamale wóolu sa bopp
   • Yokk bruit ci etiketu tàggat yi
   • Taxawaat reso bi ci noonu
3. Luy diagramu wóor?
   • Perte de formation contre époque
   • Dayo model bi ak dëggu gi
   • Probabilite biñ séentu ak bariwaay biñ seetlu
   • Tolluwaayu jàng ak dayo lote
4. Jumtukaayu etalonnage biñ seentu (ECE):
   • top-1 njubte ci model bi
   • Limu parametre yi ci model bi
   • Variance bi ci degrade yi
   • Diggante wóolu sa bopp ak jaar yoon ci biir pañe yi
5. Reseau neuronal yu xóot yi leegi dañu leen di gëna faral di wax ni:
   • Gëm sa bopp lu ëpp
   • Wóolu sa bopp
   • Kalibre bu baax ci default
   • Mënul génne ay probabilite
6. Lan moo waral model bu jaar yoon mën na soxla etalonnage?
   • Modèle yu jaar yoon duñu musa am benn njuumte ci etalonnage
   • Jaar-jaar ak etalonnage dañu wuute
   • Etalonnage amul benn njariñ ludul ci régression
   • Jaar-jaar dafay garanti probabilite yu dëggu
7. On which data is temperature scaling typically fit in the guide’s calibration workflow?
   • The final test set using classification accuracy
   • Only the training labels without model outputs
   • A held-out validation set using negative log-likelihood
   • The model’s hidden weights using mean squared error
8. Why can temperature scaling improve calibration without changing top-1 accuracy?
   • It changes the model’s predicted class labels randomly
   • It retrains the feature extractor
   • It replaces every logit with the same value
   • Dividing logits by a positive scalar changes confidence but preserves their ordering