Guide-yakabatana mibvunzo · Pakati Nhanho

Tokenizer-Yemahara Byte-Level Models Quiz

Nzwisisa mamodheru emitauro, BLT's entropy-based patches, uye kugona uye kuvharidzira tradeoffs yekubvisa yakatarwa tokeni mazwi.

Nzira dzekutungamira dzinoenderanaTokenizer Yemahara Byte Level Models
Mubvunzo 1 ye 8

Chii chinonzi tokenizer-yemahara byte-level modhi inoverengwa sezvikamu zvayo zvekupinza?

Tarisa mubvunzo wese8 mibvunzo

Mhinduro uye tsananguro zvinoonekwa paunenge uchitora bvunzo.

  1. Chii chinonzi tokenizer-yemahara byte-level modhi inoverengwa sezvikamu zvayo zvekupinza?
    • Raw bytes (UTF-8 values 0-255)
    • Mazwi-chidimbu madiki kubva pamazwi akagadziriswa
    • Mazwi ese akapatsanurwa nenzvimbo
    • Pre-computed embeddings chete
  2. Chii chinonyanya kuita chidzoreso chekudyisa mabheti akasvibira kune yakajairwa transformer?
    • Mabhayiti haamiriri mavara asiri echiRungu
    • Iwo anotevedzana anowedzera kureba, achikwidza compute mutengo
    • Pane zvakawanda zvinogoneka zvebhaiti zvakakosha kuti zvibatiswe
    • Mabhayiti anorasikirwa nerevo yese yezwi zvachose
  3. Ko iyo Byte Latent Transformer (BLT) inosarudza sei kwekuisa mabheti kuita zvigamba?
    • Nekupatsanura pane yega yega hunhu
    • Kushandisa yakagadziriswa chigamba saizi ye4 bytes
    • Kubva pane byte-level fungidziro entropy (kusava nechokwadi)
    • Randomly panguva yega yega yekupfuura
  4. Why can a fixed subword vocabulary use more tokens for some languages than others?
    • Subword tokenizers only accept ASCII
    • All languages have identical token boundaries
    • They cannot represent numerals
    • Its token inventory may reflect the language distribution in the data used to build it
  5. What does removing a fixed tokenizer guarantee about byte-level input?
    • The same byte alphabet can represent text from different scripts, while learned language ability still depends on data
    • The model understands every language without training data
    • It guarantees equal performance for all scripts
    • The model no longer needs a vocabulary or learned embeddings
  6. Ndeipi yekutanga modhi yakaratidza byte-level process asi yakashoropodzwa nekunonoka?
    • BERT
    • ByT5
    • Word2Vec
    • GPT-2
  7. What representations does BLT’s large global Transformer process?
    • Individual words from a fixed subword vocabulary only
    • Mifananidzo pixels
    • Dynamically sized byte patches
    • Robot joint-angle commands
  8. Which components surround the global Transformer in BLT?
    • A reward model and a value network
    • A speech vocoder and a spectrogram
    • A camera and a lidar sensor
    • A local byte encoder and a local byte decoder