TY  - RPRT
TI  - A Universal Trade-off Between the Model Size, Test Loss, and Training Loss of Linear Predictors
AU  - Nikhil Ghosh
AU  - Mikhail Belkin
PY  - 2023
UR  - https://arxiv.org/abs/2207.11621
ID  - 2207.11621
ER  -