quiz bu lëkkale ak tegtal · Dëgër Niveau

Quiz buy jël misaal

Saytu jaaxal bu gëna neex, rang bu lalu ci entropi, threshold mass typical_p, ak wuute gi am ci sampling top-p rang probabilité.

Yooni njiit yu jëm ciéchantillonnage
Laaj 1 de 8

Ban limu core la échantillonnage bu ñuy méngale ak token bu nekk ci kàndidaa yi?

Xoolaatal laaj bu nekk8 laaj

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

  1. Ban limu core la échantillonnage bu ñuy méngale ak token bu nekk ci kàndidaa yi?
    • ëmbiitu xibaar bi ñuy seentu (entropi) ci séddale bi
    • Dayo vocabulaire model bi
    • Position token bi ci biir frase bi
    • parametru tamperatiir bi kese
  2. naka la sampling tipik di def ci token bi gëna am wërsëg?
    • Saa yu nekk daf koy njëkka tànn
    • Mën nañu ko dindi sudee surprise bi dafa sori ci suufu valeur biñ seentu
    • Dafay yokk lu mëna am ñaari yoon
    • Dafay baña bàyyi xel ci probabilite
  3. Ban xalaat ci làkku doomi aadama yi moo waral ñu jël misaal bi?
    • Nit ñi dañuy tànn kàddu yu gëna néew
    • Nit du musa baamtu kàddu
    • Nit ñi dañuy jokkoo ci diir bu gàtt
    • Nit ñi dañuy wax ci frase yu am guddaay
  4. Lan mooy parametru tau bi ci saytu echantillonnage bi?
    • Tolluwaayu jàng ci diiru tàggat yaram
    • Limu diisaay yi ci model bi
    • Guddaayi frase bi gëna mag
    • masse probabilite cumulative biñ dajale ci ensemble tipik bi
  5. Compared with greedy decoding, what does typical sampling try to change?
    • It samples from tokens whose surprisal is closer to the distribution’s expected surprisal
    • It always selects the most probable token
    • It guarantees every continuation is factually correct
    • It removes probability scores from decoding
  6. Ci ban xeetu losisel lañuy faral di wane sampling bi ci developpeur yi?
    • Aplikaasioŋu këyitu xayma
    • Biblioteek LLM yu ubbeeku yu melni Transformatëri kanam yuy laxasu
    • Jumtukaayi soppali nataal
    • Motëri laajte ci base de done
  7. How does typical sampling rank candidate tokens differently from top-p sampling?
    • By alphabetical order of token strings
    • By token position in the prompt
    • By distance between token surprisal and expected surprisal, rather than probability rank alone
    • By the model’s training loss
  8. Which value does typical_p control?
    • The model’s vocabulary size
    • The number of transformer layers
    • The probability that the answer is factually correct
    • The target cumulative probability mass collected in the typical set