Open Models Almanac
Methods

single mention · not yet filed

Training-loss extrapolation for hyperparameter selection

The sources on this page only mention it in passing.

Also called loss extrapolation.