Guide-yakabatana mibvunzo · Pakati Nhanho

Masked Autoencoders Quiz

Ongorora MAE's yakakwirira-mask yekuvakazve mufananidzo basa, ayo asymmetric encoder-decoder dhizaini, uye bepa rakaongororwa pretraining setup.

Nzira dzekutungamira dzinoenderanaMasked Autoencoders
Mubvunzo 1 ye 8

Ndeipi basa rekutanga rekuzvitarisira muMasked Autoencoder?

Tarisa mubvunzo wese8 mibvunzo

Mhinduro uye tsananguro zvinoonekwa paunenge uchitora bvunzo.

  1. Ndeipi basa rekutanga rekuzvitarisira muMasked Autoencoder?
    • Kugadzirazve mapikseli ezvigamba aive akavanzwa
    • Kuisa mufananidzo wega wega muchikamu chakanyorwa
    • Kufanotaura furemu rinotevera mukutevedzana kwevhidhiyo
    • Kufananidza mufananidzo kune zvinyorwa zvayo
  2. Angangoita chikamu chipi chezvigamba zvemufananidzo izvo MAE yekutanga inowanzovhara kunze?
    • Zvinenge 15%
    • Zvinenge 75%
    • Zvinenge 35%
    • Zvinenge 95%
  3. Sei iyo MAE encoder ichiita chete anooneka (asina kuvharwa) zvigamba?
    • Nekuti zvigamba zvakafukidzwa hazvina ruvara runobatsira
    • Kudzivirira iyo modhi kuti isamboona mifananidzo chaiyo
    • Kuita kuti kudzidzisa kukurumidze uye kuwedzera ndangariro kushanda zvakanaka
    • Nekuti iyo decoder haigone kuverenga inomisikidzwa encodings
  4. Chii chinoitika kune decoder mushure me MAE pretraining yapera?
    • Inova iyo huru classifier
    • Yakaomeswa nechando uye inoshandiswa kugadzira mifananidzo mitsva
    • Inotsiva encoder zvachose
    • Inoraswa; iyo encoder inoshandiswa kune yakadzika mabasa
  5. Nderipi kurasikirwa basa iro MAE rinowanzo shandisa pakuvaka patsva?
    • Kurevesa squared kukanganisa pamasiki-patch pixels
    • Yambuka-entropy pamusoro pezvinyorwa zvekirasi
    • Contrastive InfoNCE kurasikirwa
    • Adversarial discriminator kurasikirwa
  6. Iyo 'asymmetric' dhizaini yeMAE inoreva chii?
    • Saizi dzemifananidzo dzakasiyana dzekupinza uye kubuda
    • Encoder inorema yakabatana neinoreruka decoder
    • Kushandisa ma dataset maviri akasiyana ekudzidzisa
    • Kuvhara hafu yeruboshwe chete yemufananidzo wese
  7. On which large image dataset did the original MAE paper pretrain its main vision models?
    • MS COCO Captions only
    • LibriSpeech
    • ImageNet-1K
    • The Netflix Prize ratings set
  8. What does the MAE decoder receive for image patches hidden from the encoder?
    • The original RGB pixels without masking
    • Only class labels from ImageNet
    • A text caption for the image
    • Learned mask tokens with positional embeddings