Maswali yanayounganishwa na mwongozo · Ngumu Kiwango

Maswali ya Uigaji wa Ruhusa ya XLNet

Hukagua urekebishaji wa vibali, umakini wa mtiririko-mbili, ujirudiaji wa Transformer-XL, na jinsi madai ya msingi yanategemea mipangilio inayoweza kulinganishwa.

Njia za mwongozo zinazohusianaUundaji wa Vibali vya Xlnet
Swali 1 ya 8

XLNet hutumia lengo gani la utayarishaji?

Hakiki kila swali8 maswali

Majibu na maelezo huonekana unapojibu maswali.

  1. XLNet hutumia lengo gani la utayarishaji?
    • Uundaji wa lugha ya vibali
    • Muundo wa lugha uliofichwa
    • Ugunduzi wa tokeni uliobadilishwa
    • Utabiri wa sentensi inayofuata
  2. Ni udhaifu gani wa BERT ambao XLNet iliundwa mahususi kushughulikia?
    • BERT ilikuwa ndogo sana
    • Tokeni ya [MASK] husababisha kutolingana kwa pretrain/fine-tune na kuchukua uhuru wa ishara iliyofichwa.
    • BERT haikuweza kushughulikia sentensi fupi
    • BERT ilikosa usimbaji nafasi
  3. Je, umakinifu wa XLNet wa mikondo miwili unatenganisha nini?
    • Tabaka za encoder na decoder
    • Grafu za mafunzo na uelekezaji
    • Mtiririko wa maudhui na mtiririko wa hoja
    • Ishara na upachikaji wa sentensi
  4. Je, XLNet inachanganya maneno katika sentensi?
    • Ndio, inapanga upya kwa nasibu ishara halisi
    • Ndio, lakini tu wakati wa kurekebisha vizuri
    • Hapana, haibadilishi uagizaji wowote hata kidogo
    • Hapana, inaruhusu agizo la uainishaji kwa kutumia vinyago vya umakini wakati wa kuweka misimamo
  5. Ni usanifu gani unaotumika kama uti wa mgongo wa XLNet kwa muktadha wa masafa marefu?
    • Kibadilishaji-XL
    • LSTM
    • Transformer ya awali ya vanilla
    • Mtandao wa mapinduzi
  6. How does XLNet combine bidirectional context with autoregressive prediction?
    • It predicts all tokens independently with no attention mask
    • It predicts in sampled factorization orders without inserting a [MASK] input token, using attention masks
    • It physically shuffles the sentence before every prediction
    • It replaces autoregressive prediction with a denoising autoencoder
  7. For a token sequence A B C, suppose a sampled factorization order predicts C before A and B. What context is available when predicting C?
    • The tokens A and B must always be provided first
    • Only the next sentence is available
    • No earlier tokens in that sampled order, although C remains at its original input position
    • The input sequence is physically reordered to C A B
  8. How does XLNet prevent the target token from leaking into its own prediction query?
    • It removes the target from the entire training sequence
    • It uses the target’s label as an input embedding
    • It replaces the token with a [MASK] symbol
    • The query stream represents the target position without its token content