83 lines
1.9 KiB
PHP
83 lines
1.9 KiB
PHP
<?php
|
|
|
|
namespace Rubix\ML\Tests\NeuralNet\Optimizers;
|
|
|
|
use Tensor\Tensor;
|
|
use Tensor\Matrix;
|
|
use Rubix\ML\NeuralNet\Parameter;
|
|
use Rubix\ML\NeuralNet\Optimizers\AdaGrad;
|
|
use Rubix\ML\NeuralNet\Optimizers\Adaptive;
|
|
use Rubix\ML\NeuralNet\Optimizers\Optimizer;
|
|
use PHPUnit\Framework\TestCase;
|
|
use Generator;
|
|
|
|
/**
|
|
* @group Optimizers
|
|
* @covers \Rubix\ML\NeuralNet\Optimizers\
|
|
*/
|
|
class AdaGradTest extends TestCase
|
|
{
|
|
/**
|
|
* @var \Rubix\ML\NeuralNet\Optimizers\AdaGrad
|
|
*/
|
|
protected $optimizer;
|
|
|
|
/**
|
|
* @before
|
|
*/
|
|
protected function setUp() : void
|
|
{
|
|
$this->optimizer = new AdaGrad(0.001);
|
|
}
|
|
|
|
/**
|
|
* @test
|
|
*/
|
|
public function build() : void
|
|
{
|
|
$this->assertInstanceOf(AdaGrad::class, $this->optimizer);
|
|
$this->assertInstanceOf(Adaptive::class, $this->optimizer);
|
|
$this->assertInstanceOf(Optimizer::class, $this->optimizer);
|
|
}
|
|
|
|
/**
|
|
* @test
|
|
* @dataProvider stepProvider
|
|
*
|
|
* @param \Rubix\ML\NeuralNet\Parameter $param
|
|
* @param \Tensor\Tensor<int|float> $gradient
|
|
* @param array[] $expected
|
|
*/
|
|
public function step(Parameter $param, Tensor $gradient, array $expected) : void
|
|
{
|
|
$this->optimizer->warm($param);
|
|
|
|
$step = $this->optimizer->step($param, $gradient);
|
|
|
|
$this->assertEquals($expected, $step->asArray());
|
|
}
|
|
|
|
/**
|
|
* @return \Generator<array>
|
|
*/
|
|
public function stepProvider() : Generator
|
|
{
|
|
yield [
|
|
new Parameter(Matrix::quick([
|
|
[0.1, 0.6, -0.4],
|
|
[0.5, 0.6, -0.4],
|
|
[0.1, 0.1, -0.7],
|
|
])),
|
|
Matrix::quick([
|
|
[0.01, 0.05, -0.02],
|
|
[-0.01, 0.02, 0.03],
|
|
[0.04, -0.01, -0.5],
|
|
]),
|
|
[
|
|
[0.001, 0.001, -0.001],
|
|
[-0.001, 0.001, 0.001],
|
|
[0.001, -0.001, -0.001],
|
|
],
|
|
];
|
|
}
|
|
}
|