A learning rate decay optimizer that reduces the learning rate by a factor of the decay parameter whenever it reaches a new floor. The number of steps needed to reach a new floor is defined by the steps parameter.
|1||rate||0.01||float||The learning rate that controls the global step size.|
|2||steps||100||int||The size of every floor in steps. i.e. the number of steps to take before applying another factor of decay.|
|3||decay||1e-3||float||The factor to decrease the learning rate at each floor.|
use Rubix\ML\NeuralNet\Optimizers\StepDecay; $optimizer = new StepDecay(0.1, 50, 1e-3);