quiz bu lëkkale ak tegtal · Dëgër Niveau

Toolformer ak Quiz Jëfandikoo Jumtukaay Jàngale Bopp

Saytu Toolformer defar wooteb boppam, segg bu sukkandiko ci ñàkk, màrker API, jumtukaay yuñ tànn ak jéego yu mujj yu baax.

Laaj 1 de 8

Lan moo wuutale Toolformer ci jëfandikoo jumtukaayi jàng?

Xoolaatal laaj bu nekk8 laaj

Sooy def quiz bi, tontu yi ak leeral yi dañuy feeñ.

  1. Lan moo wuutale Toolformer ci jëfandikoo jumtukaayi jàng?
    • Dafay defar ba noppi segg misaali tàggat jumtukaay-woote boppam
    • Dafay wéeru ci ay junniy woote jumtukaay yuñ etikete nit
    • Dafay jëfandikoo jàngat buy dooleel ci feedback nit kese
    • Dafay hard-code bepp wooteb jumtukaay ak loxo
  2. Ban tegtal la Toolformer di jëfandikoo ngir xam ndax dafay wéy di woo API kàndidaa?
    • Ndax nit ñi nangu nañu ko
    • Ndax resultaawu woote bi dafay wàññi perte model bi ci token yii
    • Ndax woote bi dafay delloo ab chaîne bu amul dara
    • Ndax woote bi du dem ci diir bu néew benn seconde
  3. Ban jumtukaay la Toolformer bi njëkk jàng jëfandikoo?
    • Generatëru nataal, sintetisatëru music, ak rendu 3D
    • Tekkikatu Python kese
    • kalkulatër, sistem QA, motëru seetlu, tekkikatu làkk, ak almanak
    • Navigateur web ak sistem fichier kese
  4. naka lañuy wane wootu jumtukaay ci biir mbind mi ci diiru liggéeyu Toolformer?
    • Ni ay chaine audio yu wuute
    • Ni nataal yuñ boole ci këyit bi
    • Ni duggal ci base de done bu biti kese
    • Ni ay token yuñ jagleel wootu API bi ak njariñam ci liiñ bi
  5. Lan mooy ñakk kattan gi ci xeetu làkk yu yaatu yi moo xiir Toolformer?
    • Dañu néew xam-xam ci xayma, xam-xam ak seetlu yu leer
    • Mënu ñu defar ay frase yu neex
    • Mënu leen gëna defar
    • Angale rek lañuy liggéey
  6. Ginaaw ñu defaree ba noppi segg misaali woote jumtukaay, lan la Toolformer di def ak ñoom?
    • sànni leen, tàmbaliwaat
    • Defar model bi ci kaw done yiñ segg te defar seen bopp
    • Daf leen di yónnee annotateur nit ñi ngir ñu etiketewaat leen
    • Dañu leen di denc ngir jàngat leen
  7. Why compare the next-token loss with and without an API result when filtering a candidate call?
    • To select the API with the longest response
    • To remove all calls that change the context
    • To retain calls whose result improves prediction of the following real text
    • To maximize the number of API calls per paragraph
  8. Once useful tool-call examples are filtered, how does Toolformer incorporate them?
    • It discards them and trains only on human ratings
    • It changes the API results into class labels
    • It retrains each external API from scratch
    • It fine-tunes on the resulting self-generated examples