Theory

Negative Overfitting Gap: The Koscak Coefficient

val_loss < train_loss. We saw it once and thought it was a bug. Then it happened again. And again. We started measuring it deliberately. We’re calling the metric something now.

⏲
In progress
Full write-up coming soon
This one needs proper framing before we publish. Coming soon.
Built on this research

Everything in this article is measured, not promised. See it running:

Talk to NOESIS liveSee the 33x benchmarkBack the research