Constant#
A learning-rate schedule that returns a fixed rate for the entire duration of training. Pair it with any optimizer when the step size should not adapt over the course of training.
Parameters#
| # | Name | Default | Type | Description |
|---|---|---|---|---|
| 1 | rate | 0.01 | float | The fixed learning rate that controls the global step size. |
Example#
use Rubix\ML\NeuralNet\Optimizers\Stochastic;
use Rubix\ML\NeuralNet\Optimizers\Schedulers\Constant;
$scheduler = new Constant(rate: 0.01);
$optimizer = new Stochastic(scheduler: $scheduler);