lrfind

			lrfind(learner[, dataiter; kwargs...]) -> LRFinderResult

Run the learning rate finder. Exponentially increases the learning rate from a very low value to a very high value and uses the losses to estimate an optimal learning rate. Return a LRFinderResult.

Keyword arguments

• nsteps = 100: maximum number of steps to run the learning rate finder for

• startlr = 1e-7: minimum learning rate

• endlr = 10: maximum learning rate

• divergefactor: stop finder early if loss goes higher than lowest loss times this factor

• estimators = [Steepest(), MinDivByTen()]: list of LREstimators