Maswali yanayounganishwa na mwongozo · Ngumu Kiwango

Maswali ya Swin Transformer

Angalia uangalizi wa dirisha lililobadilishwa, mawasiliano ya dirisha tofauti, vipengele vya daraja, kuunganisha viraka, na maombi ya kazi ya kuona katika Swin.

Njia za mwongozo zinazohusianaTransfoma ya Swin
Swali 1 ya 8

Je, 'Swin' katika Swin Transformer inawakilisha nini?

Hakiki kila swali8 maswali

Majibu na maelezo huonekana unapojibu maswali.

  1. Je, 'Swin' katika Swin Transformer inawakilisha nini?
    • Windows iliyobadilishwa
    • Dirisha Laini
    • Upepo wa anga
    • Dirisha Moja
  2. What positional information does Swin add within each local attention window?
    • Absolute word embeddings from a text tokenizer
    • A learnable relative-position bias
    • A global convolution kernel shared across all windows
    • A class-specific bounding-box regression target
  3. Swin huruhusuje habari kusonga kati ya windows tofauti?
    • Inatumia miunganisho ya mara kwa mara
    • Inafuta mipaka ya dirisha kabisa
    • Inabadilisha kizigeu cha dirisha katika tabaka zinazobadilishana
    • Inatumia dirisha moja tu la kimataifa
  4. Tabaka za kuunganisha-kiraka hufanya nini kati ya hatua za Swin?
    • Ongeza azimio na uondoe vituo
    • Kuacha mabaka bila mpangilio
    • Tafsiri viraka kwa maneno
    • Unganisha viraka vya jirani, kupunguza azimio na kuongeza chaneli
  5. Kwa nini uongozi wa Swin, pato la viwango vingi ni muhimu sana?
    • Huunganishwa vizuri katika mifumo ya ugunduzi na sehemu inayotarajia piramidi za vipengele
    • Inasaidia tu na uainishaji wa lebo moja
    • Huondoa hitaji la data iliyo na lebo
    • Inafanya mfano kuwa mdogo kuliko CNN yoyote
  6. How does Swin’s attention pattern differ from standard ViT attention?
    • Swin has no attention mechanism
    • Swin computes attention in local windows and shifts partitions between blocks; standard ViT uses global patch attention
    • ViT uses only local convolutions
    • Both always use identical global attention
  7. What scaling property does fixed-size window attention provide in Swin?
    • It is independent of image resolution
    • It becomes quadratic in channel count only
    • Attention computation grows linearly with image size for fixed window size
    • It removes all attention computation
  8. Why does shifted-window attention use a mask after the cyclic shift?
    • To force every patch to attend to every image patch
    • To remove all positional information
    • To convert window attention into a convolution
    • To prevent tokens from attending across boundaries that should remain separate windows