Ikibazo gihuza ibibazo · Hagati Urwego

Ikibazo cya ColBERT Yatinze Gukorana Ikibazo

Gerageza Ikimenyetso cya ColBERT, kodegisi yigenga, amanota ya MaxSim, akazi-ikibazo, nigihe cyo kubika cyakemuwe na compression ya ColBERTv2.

Inzira zijyanye no kuyoboraColbert Yatinze Gukorana
Ikibazo 1 Bya 8

Nigute ColBERT ihagarariye ikibazo cyangwa inyandiko?

Reba buri kibazo8 ibibazo

Ibisubizo nibisobanuro bigaragara mugihe ufata ikibazo.

  1. Nigute ColBERT ihagarariye ikibazo cyangwa inyandiko?
    • Nka impuzandengo imwe igereranijwe
    • Nka ijambo ryibanze ryibanze gusa
    • Nibimenyetso byinshi-urwego rwimiterere rwinjizamo ibice
    • Nka shusho
  2. Nibihe bikorwa byo gutanga amanota ColBERT ikoresha yitwa?
    • MaxSim
    • Kuzunguruka
    • Softmax cross-entropy
    • Gushakisha ibiti
  3. What query-time work does ColBERT avoid by precomputing document token vectors?
    • All query encoding and all index access
    • Every comparison between a query token and document token
    • The need to store document representations
    • Joint transformer encoding of every query-document pair, while query encoding and similarity search still run
  4. Niki 'gutinda' muri 'gutinda gukorana' bivuga iki?
    • Icyitegererezo cyatojwe bitinze
    • Imikoranire hagati yikibazo ninyandiko zerekana koherezwa ku ntambwe yo gutanga amanota nyuma ya kodegisi yigenga
    • Ibibazo bigera nijoro
    • Inyandiko zongeweho nyuma yo koherezwa
  5. Ni ikihe kibazo kijyanye no kugarura inshuro imwe ColBERT ikemura?
    • Biratinda cyane kurutonde
    • Ntishobora gukora kuri GPUs
    • Irasaba amakuru yanditse kuri buri kibazo
    • Gucomeka igice cyose muri vector imwe itakaza ibisobanuro birambuye
  6. Niki ColBERTv2 yateye imbere cyane kurenza umwimerere?
    • Yakuyeho ibikenewe byose
    • Yahinduye kuva BERT yerekeza kuri sisitemu ishingiye kumategeko
    • Yongeyeho compression isigaye kugirango igabanye cyane ingano yubunini
    • Yakuyeho intambwe ya MaxSim
  7. A two-token ColBERT query has best document-token similarities of 0.8 and 0.3. What MaxSim score results?
    • 0.8, retaining only the single largest value across the whole query
    • 1.1, the sum of the two per-query-token maxima
    • 0.55, averaging instead of summing the maxima
    • 0.24, multiplying the two maxima
  8. Why can a ColBERT document index require more storage than a single-vector dense index?
    • It retains multiple token-level representations for each document
    • It never precomputes any document features
    • It must store all transformer gradients for each document
    • It removes every document token before indexing