TY  - RPRT
TI  - Why you don't overfit, and don't need Bayes if you only train for one epoch
AU  - Laurence Aitchison
PY  - 2024
UR  - https://arxiv.org/abs/2411.14478
ID  - 2411.14478
ER  -