Guide-yakabatana mibvunzo · Pakati Nhanho

Musanganiswa Wezvakadzama Quiz

Yedza MoD's yakadzidza tokeni router, yakagadziriswa yepamusoro-k huwandu, yasara nzira yeakasvetuka tokens, static zvimiro, uye causal nzira kuchengetedza.

Nzira dzekutungamira dzinoenderanaMusanganiswa Wezvakadzama
Mubvunzo 1 ye 8

Chii chinonzi Musanganiswa Wezvakadzama unosiyana pane tokens?

Tarisa mubvunzo wese8 mibvunzo

Mhinduro uye tsananguro zvinoonekwa paunenge uchitora bvunzo.

  1. Chii chinonzi Musanganiswa Wezvakadzama unosiyana pane tokens?
    • Yakawanda sei compute (api akaturikidzana) tokeni imwe neimwe inogamuchira
    • Izwi rinoshandiswa
    • Iyo dataset yekudzidziswa
    • Mutauro wekubuda
  2. MoD inochengeta sei bhajeti rayo rekufungidzira?
    • Nekungodonhedza hafu yezviyereso zvemuenzaniso
    • Nekugadzirisa chikamu chepamusoro-k chikamu chematokeni pane imwe neimwe block
    • Nokushandisa muenzaniso wakasiyana pamutsara wega wega
    • Nokudzima pfungwa zvachose
  3. Chii chinoitika kune tokens iyo router haina sarudza pane yakapihwa block?
    • Vanodzimwa kubva pakutevedzana
    • Vanotumirwa kune imwe muenzaniso
    • Ivo vanosvetuka block kuburikidza neinosara yekubatanidza, vachipfuura nepasina kuchinjika
    • Vanoshandurwa kuita mifananidzo
  4. Which lab published the Mixture of Depths technique?
    • OpenAI muna 2019
    • Meta muna 2015
    • Microsoft muna 2012
    • Google DeepMind muna 2024
  5. Chii chinopa kusanganisa MoD neMusanganiswa-we-Nyanzvi?
    • MoDE, iyo inofamba pane ese kudzika uye nyanzvi sarudzo
    • A modhi isina parameters
    • Yakachena convolutional network
    • A tokenizer-chete system
  6. Sei yakagadziriswa yepamusoro-k inosarudzwa pamusoro pechikumbaridzo chechiratidzo muMoD?
    • Inoita iyo modhi inobuda zvisina tsarukano zvinyorwa
    • Inochengeta tensor maumbirwo akamira, ayo anoshanda pane Hardware
    • Inobvisa kudiwa kwekudzidziswa
    • Inowedzera hukuru hwemashoko
  7. What does the learned router do at a Mixture-of-Depths block?
    • Selects a new model family for every prompt
    • Predicts labels for the next training batch
    • Scores tokens so the block can select which ones receive full attention and MLP computation
    • Changes the vocabulary size at inference
  8. Why does causal generation in MoD use auxiliary predictors for routing?
    • To let a model read the entire answer before generating
    • To remove the fixed top-k compute budget
    • To convert the sequence into an image
    • To make routing decisions without looking at future tokens