Idanwo ti o sopọ mọ itọsọna · Alabọde Ipele

Idanwo Ti o kere julọ-si-julọ julọ

Test how Least-to-Most decomposes difficult questions and uses sequential solutions to improve easy-to-hard generalization.

Jẹmọ itọsọna awọn ọnaTi o kere julọ si Itọkasi julọ
Ibeere 1 ti 8

Kini awọn ipele akọkọ meji ti o kere julọ-si-Pupọ?

Awotẹlẹ gbogbo ibeere8 ibeere

Awọn idahun ati awọn alaye han bi o ṣe n mu ibeere naa.

  1. Kini awọn ipele akọkọ meji ti o kere julọ-si-Pupọ?
    • Pa iṣoro naa sinu awọn ibeere, lẹhinna yanju wọn ni ibere
    • Kọ ẹkọ lẹhinna dara-tune
    • Encode lẹhinna yan koodu
    • Ayẹwo ọpọlọpọ awọn idahun lẹhinna dibo
  2. Bawo ni Kere-si-Pupọ ṣe yato si imuduro pq-ti-ero ti o ṣe deede?
    • Ko lo apẹẹrẹ rara
    • O ṣe afihan iṣoro naa ni gbangba ati pe o jẹ ifunni ipin kọọkan sinu igbesẹ ti n tẹle
    • O nilo atunṣe awoṣe daradara
    • O ṣiṣẹ nikan lori awọn aworan
  3. What kind of task does the SCAN benchmark use to test compositional generalization?
    • Classifying satellite images of fishing vessels
    • Predicting fuel use from ship weather data
    • Interpreting natural-language commands that combine primitive actions
    • Aligning speech frames with handwritten characters
  4. Kini idi ti aṣẹ ti awọn ibeere ṣe pataki?
    • Kii ṣe; ibere ni ID
    • Awọn ibeere nigbamii gbọdọ wa ni akọkọ
    • Ibere nikan ni ipa lori akoonu
    • Ibeere kọọkan yẹ ki o dale lori awọn ti a ti yanju tẹlẹ ṣaaju rẹ
  5. Bawo ni awọn idahun agbedemeji ṣe lo lakoko ipele ipinnu?
    • Wọn ti wa ni afikun si ọrọ-ọrọ ki awọn igbesẹ nigbamii le kọ lori wọn
    • Wọn ti sọnu lẹhin igbesẹ kọọkan
    • Wọn ti wa ni rán si lọtọ awoṣe
    • Wọn jẹ aropin si nọmba kan
  6. Kini eewu bọtini kan ninu ẹwọn Kere-si-Pupọ?
    • Ko le lo awọn apẹẹrẹ
    • Aṣiṣe ni kutukutu ti ko tọ le ba gbogbo awọn igbesẹ isalẹ jẹ
    • O ṣiṣẹ nikan ni ede kan
    • Nigbagbogbo o nilo iṣupọ GPU kan
  7. What accuracy did the paper report for code-davinci-002 on the SCAN length split using Least-to-Most with 14 exemplars?
    • About 16%
    • Gangan 50%
    • At least 99% on the reported setup
    • Zero, because no demonstrations were used
  8. Besides SCAN, which task areas did the Least-to-Most paper also evaluate?
    • Wildlife re-identification and acoustic classification
    • Ship routing and port scheduling
    • Image outpainting and watermark detection
    • Symbolic manipulation and math reasoning