Maswali yanayounganishwa na mwongozo · Ngumu Kiwango

Maswali ya Sampuli ya Kawaida

Hukagua mshangao wa kawaida, cheo kulingana na entropy, kizingiti cha kawaida cha_p, na tofauti kutoka kwa sampuli za nafasi ya juu ya uwezekano.

Njia za mwongozo zinazohusianaSampuli ya Kawaida
Swali 1 ya 8

Sampuli ya kawaida inalinganisha kiasi gani cha msingi dhidi ya kila toni ya mtahiniwa?

Hakiki kila swali8 maswali

Majibu na maelezo huonekana unapojibu maswali.

  1. Sampuli ya kawaida inalinganisha kiasi gani cha msingi dhidi ya kila toni ya mtahiniwa?
    • Maudhui ya habari yanayotarajiwa (entropy) ya usambazaji
    • Saizi ya jumla ya msamiati wa mtindo
    • Nafasi ya ishara katika sentensi
    • Kigezo cha joto pekee
  2. Sampuli za kawaida hushughulikiaje ishara moja inayowezekana zaidi?
    • Daima huchagua kwanza
    • Inaweza kuitenga ikiwa mshangao wake uko chini ya thamani inayotarajiwa
    • Inaongeza uwezekano wake mara mbili
    • Inapuuza uwezekano kabisa
  3. Sampuli za kawaida zilichochewa na wazo gani kuhusu lugha ya binadamu?
    • Wanadamu daima huchagua maneno adimu iwezekanavyo
    • Wanadamu kamwe hawarudii maneno
    • Wanadamu huwasiliana karibu na kiwango cha habari thabiti
    • Wanadamu huzungumza kwa sentensi zenye urefu usiobadilika
  4. Je, parameta ya tau katika udhibiti wa sampuli ya kawaida hutumika nini?
    • Kiwango cha mafunzo wakati wa mafunzo
    • Idadi ya tabaka katika mfano
    • Urefu wa juu wa sentensi
    • Wingi wa uwezekano unaokusanywa katika seti ya kawaida
  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. Ni aina gani ya programu ambayo sampuli za kawaida huonyeshwa kwa wasanidi programu?
    • Programu za lahajedwali
    • Maktaba za uelekezaji za LLM za chanzo huria kama vile Hugging Face Transformers
    • Zana za kuhariri picha
    • Injini za hoja za hifadhidata
  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