Échantillonnage négatif ak bruit

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

1. Ban jafe-jafe la échantillonnage négatif ak NCE njëkka saafara?
   • Njëg yu bari yi ngir xayma softmax bu mat ci kaw bayaal bu yaatu lool
   • Overfitting ci done yu ndaw
   • Gradient yuy réer ci reso yuy baaxoo
   • Ñàkum done yuñ etikete ngir jàng buñ yor
2. naka la ko estimaasioŋ buy wuutale bruit di defaree liggéeyu jàng bi?
   • Regression ci kaw ay mébet yu wéy
   • Ni xaaj binär diggante misaali done dëgg ak misaali bruit
   • Ni mbooloo mu am poñ yuñ etiketewul
   • Ni jàngat buy dooleel ak neexal
3. Ban model moo siiwal sampling négatif ngir jàng embeddings baat?
   • Bert
   • Resnet
   • xeetu gram bu word2vec
   • GPT-2
4. Lan moo wuutale échantillonnage négatif ak NCE bu mat?
   • Dafay yokk ay terme normalisasioŋ ngir gëna jubal
   • Dafay laaj vocabulaire bi yépp ci jéego bu nekk
   • Dafay jëfandikoo ñàkka régression ci barabu sigmoid
   • Dafay dindi NCE ci wàllu normalisasioŋ, gëna gaaw waaye du estimatëru densité bu méngoo
5. Lan moo waral ñuy faral di jël negatif yi ci distribution unigram bi yéeg ba 3/4 doole?
   • ngir yemale njeexitalu baat yu bari ak yu bariwul luñuy jël misaal ci bariwaay
   • ngir baña bàyyi xel ci kàddu yu bariwul
   • Ngir baat yépp di méngoo
   • Dindi kàddu yuy tere tàggat yaram
6. For a large vocabulary, how can sampled negative pairs reduce work compared with evaluating every word in a full softmax?
   • It computes an exact probability for every vocabulary item
   • The update scores a positive pair and a limited set of sampled negatives instead of normalizing over the whole vocabulary
   • It eliminates the need to score the positive pair
   • It sorts all words by frequency on each step
7. In skip-gram negative sampling, what is treated as the positive training pair?
   • Two words sampled independently from the noise distribution
   • A word paired with an empty context
   • A word and a context word observed together in the corpus
   • A validation label paired with a model checkpoint
8. Why is negative sampling not generally interpreted as estimating a normalized language-model probability?
   • It always sums over the full vocabulary
   • It requires exact maximum-likelihood normalization
   • It never uses negative examples
   • Its simplified binary objective drops NCE normalization terms and targets useful embeddings instead