Double Isọkale Ìdánwò

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

1. Nibo ni aṣiṣe idanwo ṣe deede tente oke ni ilọ-isalẹ ilọpo meji?
   • Ni ẹnu-ọna interpolation nibiti awoṣe kan baamu gbogbo data ikẹkọ
   • Ni ibẹrẹ ikẹkọ pẹlu awọn paramita diẹ
   • Nikan lẹhin ọkẹ àìmọye ti paramita
   • Aṣiṣe idanwo ko ga ju; o ṣubu monotonically
2. Kini yoo ṣẹlẹ lati ṣe idanwo aṣiṣe bi awoṣe ṣe di iwọn apọju pupọ?
   • O ga soke lainidii nitori overfitting
   • O le dinku lẹẹkansi, nigbagbogbo ni isalẹ awọn kilasika kere
   • O duro gangan alapin
   • O di ID ati airotẹlẹ
3. In which setting does the cited implicit-bias analysis characterize a minimum-norm solution?
   • Any neural network trained with gradient descent
   • Any model that reaches zero training error
   • A single-hidden-layer linear network under stated gradient-flow and initialization assumptions
   • A model trained without an optimization objective
4. 'Epoch-wise' ilọpo ilọpo meji n tọka si ipa ti o farahan bi iṣẹ ti kini?
   • Awọn nọmba ti GPUs lo
   • Iṣeto-oṣuwọn ẹkọ nikan
   • Awọn iwọn ti awọn igbeyewo ṣeto
   • Akoko ikẹkọ (nọmba ti epochs)
5. Eyi ti kilasika agutan ni ilopo ayalu ipenija?
   • Ibajẹ-iyatọ U-sókè iṣowo
   • Itankalẹ afẹyinti
   • Sitokasitik gradient iran
   • Ofin pq
6. Tani o ṣe iranlọwọ fun olokiki ati ṣe agbekalẹ ọna ilọpo ilọpo meji ni ọdun 2019?
   • Newton ati Leibniz
   • Belkin, Hsu, Ma, ati Mandal
   • Turing ati Ijo
   • Nikan Anonymous Kaggle oludije
7. Which model families did OpenAI report studying in its deep double-descent experiments?
   • Only decision trees and linear regression
   • Only recurrent speech recognizers
   • Convolutional neural networks, ResNets, and transformers
   • Only graphical game generators
8. Besides model size and training duration, which axis can also show double-descent behavior?
   • Font size in a report
   • The number of words in a product name
   • The screen resolution of a monitor
   • Training dataset size