Kedis ku xidhan hagaha · Adag Heerka

Kediska Muunada ee Caadiga ah

Wuxuu hubiyaa la yaabka caadiga ah, darajada ku salaysan entropy, heerka caadiga ah ee caadiga ah, iyo kala duwanaanshaha jaantuska sare-p ee suurtogalka ah.

Wadooyinka haga ee la xidhiidhaSaamaynta caadiga ah
Su'aal 1 ee 8

Waa maxay tirada xudunta u ah muunad kasta oo la barbar dhigo musharraxiinta?

Horudhac su'aal kasta8 su'aalo

Jawaabaha iyo sharraxaadaha waxay soo baxayaan markaad qaadanayso kediska.

  1. Waa maxay tirada xudunta u ah muunad kasta oo la barbar dhigo musharraxiinta?
    • Macluumaadka la filayo (entropy) qaybinta
    • Qaabka guud ahaan cabbirka erayada
    • Booska calaamada jumlada
    • Halbeegga heerkulka oo keliya
  2. Sidee buu muunada caadiga ah u daaweeyaa calaamadda keliya ee suurtogalka ah?
    • Had iyo jeer waxay doorataa marka hore
    • Waxa laga yaabaa inay ka saarto haddii yaabkeedu aad uga hooseeyo qiimaha la filayo
    • Waxay labanlaabtaa ixtimaalka
    • Waxay iska indhatiraysaa ixtimaalka gabi ahaanba
  3. Muunad-qaadista caadiga ah waxaa dhiirigeliyay fikraddee ku saabsan luqadda aadanaha?
    • Bani'aadamku had iyo jeer waxay doortaan erayada suurtagalka ah ee dhifka ah
    • Aadamuhu weligood kuma celceliyaan erayada
    • Bini'aadanku waxay ku wada xidhiidhaan ku dhawaad qiyaas ahaan xog joogto ah
    • Aadanuhu waxay ku hadlaan weedho dhererkoodu go'an yahay
  4. Waa maxay cabbirka tau ee kantaroolka muunada caadiga ah?
    • Heerka waxbarashada inta lagu jiro tababarka
    • Tirada lakabyada qaabka
    • Dhererka jumlada ugu badan
    • Tirada ixtimaalka isugeynta ayaa lagu soo ururiyay qaabka caadiga ah
  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. Waa maxay nooca software-ka caadiga ah ee muunad caadi u ah soo-saarayaasha?
    • Codsiyada xaashida
    • Isha furan ee laybareeriyada LLM sida Hugging Face Transformers
    • Qalabka tafatirka sawirka
    • Matoorada weydiimaha database
  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?
    • Cabbirka eraybixinta moodelka
    • The number of transformer layers
    • The probability that the answer is factually correct
    • The target cumulative probability mass collected in the typical set