Open Models Almanac
Methods

single mention ยท not yet filed

Parabola fit in log10 learning rate space to find optimal learning rate

The sources on this page only mention it in passing.