cavis/libnd4j/tests_cpu/layers_tests/RNGTests.cpp

1004 lines
33 KiB
C++
Raw Normal View History

2019-06-06 14:21:15 +02:00
/*******************************************************************************
* Copyright (c) 2015-2018 Skymind, Inc.
* Copyright (c) 2019 Konduit K.K.
2019-06-06 14:21:15 +02:00
*
* This program and the accompanying materials are made available under the
* terms of the Apache License, Version 2.0 which is available at
* https://www.apache.org/licenses/LICENSE-2.0.
*
* Unless required by applicable law or agreed to in writing, software
* distributed under the License is distributed on an "AS IS" BASIS, WITHOUT
* WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the
* License for the specific language governing permissions and limitations
* under the License.
*
* SPDX-License-Identifier: Apache-2.0
******************************************************************************/
//
// @author raver119@gmail.com
//
#include "testlayers.h"
#include <chrono>
#include <NDArray.h>
#include <helpers/RandomLauncher.h>
#include <ops/declarable/LegacyRandomOp.h>
#include <ops/declarable/CustomOperations.h>
using namespace nd4j;
class RNGTests : public testing::Test {
private:
//Nd4jLong *_bufferA;
//Nd4jLong *_bufferB;
public:
long _seed = 119L;
//nd4j::random::RandomBuffer *_rngA;
//nd4j::random::RandomBuffer *_rngB;
nd4j::graph::RandomGenerator _rngA;
nd4j::graph::RandomGenerator _rngB;
NDArray* nexp0 = NDArrayFactory::create_<float>('c', {10, 10});
NDArray* nexp1 = NDArrayFactory::create_<float>('c', {10, 10});
NDArray* nexp2 = NDArrayFactory::create_<float>('c', {10, 10});
RNGTests() {
//_bufferA = new Nd4jLong[100000];
//_bufferB = new Nd4jLong[100000];
//_rngA = (nd4j::random::RandomBuffer *) initRandom(nullptr, _seed, 100000, (Nd4jPointer) _bufferA);
//_rngB = (nd4j::random::RandomBuffer *) initRandom(nullptr, _seed, 100000, (Nd4jPointer) _bufferB);
2019-06-06 14:21:15 +02:00
_rngA.setStates(_seed, _seed);
_rngB.setStates(_seed, _seed);
nexp0->assign(-1.0f);
nexp1->assign(-2.0f);
nexp2->assign(-3.0f);
}
~RNGTests() {
//destroyRandom(_rngA);
//destroyRandom(_rngB);
2019-06-06 14:21:15 +02:00
//delete[] _bufferA;
//delete[] _bufferB;
delete nexp0;
delete nexp1;
delete nexp2;
}
};
TEST_F(RNGTests, TestSeeds_1) {
RandomGenerator generator(123L, 456L);
ASSERT_EQ(123, generator.rootState());
ASSERT_EQ(456, generator.nodeState());
Nd4jPointer ptr = malloc(sizeof(RandomGenerator));
memcpy(ptr, &generator, sizeof(RandomGenerator));
auto cast = reinterpret_cast<RandomGenerator*>(ptr);
ASSERT_EQ(123, cast->rootState());
ASSERT_EQ(456, cast->nodeState());
free(ptr);
}
TEST_F(RNGTests, TestSeeds_2) {
RandomGenerator generator(12, 13);
generator.setStates(123L, 456L);
ASSERT_EQ(123, generator.rootState());
ASSERT_EQ(456, generator.nodeState());
}
TEST_F(RNGTests, Test_Dropout_1) {
auto x0 = NDArrayFactory::create<float>('c', {10, 10});
auto x1 = NDArrayFactory::create<float>('c', {10, 10});
x0.linspace(1);
x1.linspace(1);
float prob[] = {0.5f};
//x0.applyRandom(random::DropOut, _rngA, nullptr, &x0, prob);
//x1.applyRandom(random::DropOut, _rngB, nullptr, &x1, prob);
RandomLauncher::applyDropOut(LaunchContext::defaultContext(), _rngA, &x0, 0.5);
RandomLauncher::applyDropOut(LaunchContext::defaultContext(), _rngB, &x1, 0.5);
2019-06-06 14:21:15 +02:00
ASSERT_TRUE(x0.equalsTo(&x1));
//x0.printIndexedBuffer("Dropout");
// this check is required to ensure we're calling wrong signature
ASSERT_FALSE(x0.equalsTo(nexp0));
ASSERT_FALSE(x0.equalsTo(nexp1));
ASSERT_FALSE(x0.equalsTo(nexp2));
}
TEST_F(RNGTests, Test_DropoutInverted_1) {
auto x0 = NDArrayFactory::create<float>('c', {10, 10});
auto x1 = NDArrayFactory::create<float>('c', {10, 10});
x0.linspace(1);
x1.linspace(1);
float prob[] = {0.5f};
//x0.template applyRandom<randomOps::DropOutInverted<float>>(_rngA, nullptr, &x0, prob);
//x1.template applyRandom<randomOps::DropOutInverted<float>>(_rngB, nullptr, &x1, prob);
RandomLauncher::applyInvertedDropOut(LaunchContext::defaultContext(), _rngA, &x0, 0.5);
RandomLauncher::applyInvertedDropOut(LaunchContext::defaultContext(), _rngB, &x1, 0.5);
2019-06-06 14:21:15 +02:00
ASSERT_TRUE(x0.equalsTo(&x1));
//x0.printIndexedBuffer("DropoutInverted");
// this check is required to ensure we're calling wrong signature
ASSERT_FALSE(x0.equalsTo(nexp0));
ASSERT_FALSE(x0.equalsTo(nexp1));
ASSERT_FALSE(x0.equalsTo(nexp2));
}
TEST_F(RNGTests, Test_Launcher_1) {
auto x0 = NDArrayFactory::create<float>('c', {10, 10});
auto x1 = NDArrayFactory::create<float>('c', {10, 10});
RandomLauncher::applyDropOut(LaunchContext::defaultContext(), _rngA, &x0, 0.5f);
RandomLauncher::applyDropOut(LaunchContext::defaultContext(), _rngB, &x1, 0.5f);
2019-06-06 14:21:15 +02:00
ASSERT_TRUE(x0.equalsTo(&x1));
ASSERT_FALSE(x0.equalsTo(nexp0));
ASSERT_FALSE(x0.equalsTo(nexp1));
ASSERT_FALSE(x0.equalsTo(nexp2));
}
TEST_F(RNGTests, Test_Launcher_2) {
auto x0 = NDArrayFactory::create<float>('c', {10, 10});
auto x1 = NDArrayFactory::create<float>('c', {10, 10});
RandomLauncher::applyInvertedDropOut(LaunchContext::defaultContext(), _rngA, &x0, 0.5f);
RandomLauncher::applyInvertedDropOut(LaunchContext::defaultContext(), _rngB, &x1, 0.5f);
2019-06-06 14:21:15 +02:00
ASSERT_TRUE(x0.equalsTo(&x1));
ASSERT_FALSE(x0.equalsTo(nexp0));
ASSERT_FALSE(x0.equalsTo(nexp1));
ASSERT_FALSE(x0.equalsTo(nexp2));
}
TEST_F(RNGTests, Test_Launcher_3) {
auto x0 = NDArrayFactory::create<float>('c', {10, 10});
auto x1 = NDArrayFactory::create<float>('c', {10, 10});
RandomLauncher::applyAlphaDropOut(LaunchContext::defaultContext(), _rngA, &x0, 0.5f, 0.2f, 0.1f, 0.3f);
RandomLauncher::applyAlphaDropOut(LaunchContext::defaultContext(), _rngB, &x1, 0.5f, 0.2f, 0.1f, 0.3f);
2019-06-06 14:21:15 +02:00
//x1.printIndexedBuffer("x1");
ASSERT_TRUE(x0.equalsTo(&x1));
ASSERT_FALSE(x0.equalsTo(nexp0));
ASSERT_FALSE(x0.equalsTo(nexp1));
ASSERT_FALSE(x0.equalsTo(nexp2));
}
TEST_F(RNGTests, Test_Uniform_1) {
auto x0 = NDArrayFactory::create<float>('c', {10, 10});
auto x1 = NDArrayFactory::create<float>('c', {10, 10});
RandomLauncher::fillUniform(LaunchContext::defaultContext(), _rngA, &x0, 1.0f, 2.0f);
RandomLauncher::fillUniform(LaunchContext::defaultContext(), _rngB, &x1, 1.0f, 2.0f);
2019-06-06 14:21:15 +02:00
ASSERT_TRUE(x0.equalsTo(&x1));
ASSERT_FALSE(x0.equalsTo(nexp0));
ASSERT_FALSE(x0.equalsTo(nexp1));
ASSERT_FALSE(x0.equalsTo(nexp2));
for (int e = 0; e < x0.lengthOf(); e++) {
float v = x0.e<float>(e);
ASSERT_TRUE(v >= 1.0f && v <= 2.0f);
}
}
TEST_F(RNGTests, Test_Uniform_3) {
auto x0 = NDArrayFactory::create<double>('c', {1000000});
RandomLauncher::fillUniform(LaunchContext::defaultContext(), _rngA, &x0, 1.0f, 2.0f);
2019-06-06 14:21:15 +02:00
for (int e = 0; e < x0.lengthOf(); e++) {
auto v = x0.t<double>(e);
ASSERT_TRUE(v >= 1.0 && v <= 2.0);
}
}
TEST_F(RNGTests, Test_Bernoulli_1) {
auto x0 = NDArrayFactory::create<float>('c', {10, 10});
auto x1 = NDArrayFactory::create<float>('c', {10, 10});
RandomLauncher::fillBernoulli(LaunchContext::defaultContext(), _rngA, &x0, 1.0f);
RandomLauncher::fillBernoulli(LaunchContext::defaultContext(), _rngB, &x1, 1.0f);
2019-06-06 14:21:15 +02:00
ASSERT_TRUE(x0.equalsTo(&x1));
ASSERT_FALSE(x0.equalsTo(nexp0));
ASSERT_FALSE(x0.equalsTo(nexp1));
ASSERT_FALSE(x0.equalsTo(nexp2));
}
TEST_F(RNGTests, Test_Gaussian_1) {
auto x0 = NDArrayFactory::create<float>('c', {10, 10});
auto x1 = NDArrayFactory::create<float>('c', {10, 10});
RandomLauncher::fillGaussian(LaunchContext::defaultContext(), _rngA, &x0, 1.0f, 2.0f);
RandomLauncher::fillGaussian(LaunchContext::defaultContext(), _rngB, &x1, 1.0f, 2.0f);
2019-06-06 14:21:15 +02:00
//x0.printIndexedBuffer("x0");
//x1.printIndexedBuffer("x1");
ASSERT_TRUE(x0.equalsTo(&x1));
ASSERT_FALSE(x0.equalsTo(nexp0));
ASSERT_FALSE(x0.equalsTo(nexp1));
ASSERT_FALSE(x0.equalsTo(nexp2));
}
TEST_F(RNGTests, Test_Gaussian_21) {
auto x0 = NDArrayFactory::create<float>('c', {10, 10});
auto x1 = NDArrayFactory::create<float>('c', {10, 10});
RandomLauncher::fillGaussian(LaunchContext::defaultContext(), _rngA, &x0, 0.0f, 1.0f);
RandomLauncher::fillGaussian(LaunchContext::defaultContext(), _rngB, &x1, 0.0f, 1.0f);
2019-06-06 14:21:15 +02:00
// x0.printIndexedBuffer("x0");
// x1.printIndexedBuffer("x1");
2019-06-06 14:21:15 +02:00
ASSERT_TRUE(x0.equalsTo(&x1));
ASSERT_FALSE(x0.equalsTo(nexp0));
ASSERT_FALSE(x0.equalsTo(nexp1));
ASSERT_FALSE(x0.equalsTo(nexp2));
nd4j::ops::moments op;
auto result = op.execute({&x0}, {}, {});
//x0.printIndexedBuffer("X0 Normal");
//x1.printIndexedBuffer("X1 Normal");
ASSERT_TRUE(result->status() == Status::OK());
auto mean = result->at(0);
auto variance = result->at(1);
// mean->printIndexedBuffer("Mean");
// variance->printIndexedBuffer("Variance");
ASSERT_NEAR(nd4j::math::nd4j_abs(mean->e<float>(0)), 0.f, 0.2f);
ASSERT_NEAR(variance->e<float>(0), 1.0f, 0.2f);
delete result;
}
#ifdef DEBUG_BUILD
2019-06-06 14:21:15 +02:00
TEST_F(RNGTests, Test_Gaussian_22) {
auto x0 = NDArrayFactory::create<float>('c', {10000, 1000});
auto x1 = NDArrayFactory::create<float>('c', {10000, 1000});
[WIP] More of CUDA operations (#69) * initial commit Signed-off-by: raver119 <raver119@gmail.com> * - gruCell_bp further Signed-off-by: Yurii <yurii@skymind.io> * - further work on gruCell_bp Signed-off-by: Yurii <yurii@skymind.io> * Inverse matrix cublas implementation. Partial working revision. * Separation of segment ops helpers. Max separation. * Separated segment_min ops. * Separation of segment_mean/sum/prod/sqrtN ops heleprs. * Fixed diagonal processing with LUP decomposition. * Modified inversion approach using current state of LU decomposition. * Implementation of matrix_inverse op with cuda kernels. Working revision. * Implemented sequence_mask cuda helper. Eliminated waste printf with matrix_inverse implementation. Added proper tests. * - further work on gruCell_bp (ff/cuda) Signed-off-by: Yurii <yurii@skymind.io> * comment one test for gruCell_bp Signed-off-by: Yurii <yurii@skymind.io> * - provide cuda static_rnn Signed-off-by: Yurii <yurii@skymind.io> * Refactored random_shuffle op to use new random generator. * Refactored random_shuffle op helper. * Fixed debug tests with random ops tests. * Implement random_shuffle op cuda kernel helper and tests. * - provide cuda scatter_update Signed-off-by: Yurii <yurii@skymind.io> * Implementation of random_shuffle for linear case with cuda kernels and tests. * Implemented random_shuffle with cuda kernels. Final revision. * - finally gruCell_bp is completed Signed-off-by: Yurii <yurii@skymind.io> * Dropout op cuda helper implementation. * Implemented dropout_bp cuda helper. * Implemented alpha_dropout_bp with cuda kernel helpers. * Refactored helper. * Implementation of suppresion helper with cuda kernels. * - provide cpu code fot hsvToRgb, rgbToHsv, adjustHue Signed-off-by: Yurii <yurii@skymind.io> * Using sort by value method. * Implementation of image.non_max_suppression op cuda-based helper. * - correcting and testing adjust_hue, adjust_saturation cpu/cuda code Signed-off-by: Yurii <yurii@skymind.io> * Added cuda device prefixes to declarations. * Implementation of hashcode op with cuda helper. Initital revision. * rnn cu impl removed Signed-off-by: raver119 <raver119@gmail.com>
2019-07-20 07:58:44 +02:00
RandomLauncher::fillGaussian(nd4j::LaunchContext::defaultContext(), _rngA, &x0, 0.0f, 1.0f);
RandomLauncher::fillGaussian(LaunchContext::defaultContext(), _rngB, &x1, 0.0f, 1.0f);
2019-06-06 14:21:15 +02:00
//x0.printIndexedBuffer("x0");
//x1.printIndexedBuffer("x1");
ASSERT_TRUE(x0.equalsTo(&x1));
ASSERT_FALSE(x0.equalsTo(nexp0));
ASSERT_FALSE(x0.equalsTo(nexp1));
ASSERT_FALSE(x0.equalsTo(nexp2));
nd4j::ops::moments op;
auto result = op.execute({&x0}, {}, {});
//x0.printIndexedBuffer("X0 Normal");
//x1.printIndexedBuffer("X1 Normal");
ASSERT_TRUE(result->status() == Status::OK());
auto mean0 = result->at(0);
auto variance0 = result->at(1);
//mean0->printIndexedBuffer("Mean");
//variance0->printIndexedBuffer("Variance");
ASSERT_NEAR(nd4j::math::nd4j_abs(mean0->e<float>(0)), 0.f, 1.0e-3f);
ASSERT_NEAR(variance0->e<float>(0), 1.0f, 1.e-3f);
delete result;
}
TEST_F(RNGTests, Test_Gaussian_3) {
auto x0 = NDArrayFactory::create<double>('c', {10000000});
[WIP] More of CUDA operations (#69) * initial commit Signed-off-by: raver119 <raver119@gmail.com> * - gruCell_bp further Signed-off-by: Yurii <yurii@skymind.io> * - further work on gruCell_bp Signed-off-by: Yurii <yurii@skymind.io> * Inverse matrix cublas implementation. Partial working revision. * Separation of segment ops helpers. Max separation. * Separated segment_min ops. * Separation of segment_mean/sum/prod/sqrtN ops heleprs. * Fixed diagonal processing with LUP decomposition. * Modified inversion approach using current state of LU decomposition. * Implementation of matrix_inverse op with cuda kernels. Working revision. * Implemented sequence_mask cuda helper. Eliminated waste printf with matrix_inverse implementation. Added proper tests. * - further work on gruCell_bp (ff/cuda) Signed-off-by: Yurii <yurii@skymind.io> * comment one test for gruCell_bp Signed-off-by: Yurii <yurii@skymind.io> * - provide cuda static_rnn Signed-off-by: Yurii <yurii@skymind.io> * Refactored random_shuffle op to use new random generator. * Refactored random_shuffle op helper. * Fixed debug tests with random ops tests. * Implement random_shuffle op cuda kernel helper and tests. * - provide cuda scatter_update Signed-off-by: Yurii <yurii@skymind.io> * Implementation of random_shuffle for linear case with cuda kernels and tests. * Implemented random_shuffle with cuda kernels. Final revision. * - finally gruCell_bp is completed Signed-off-by: Yurii <yurii@skymind.io> * Dropout op cuda helper implementation. * Implemented dropout_bp cuda helper. * Implemented alpha_dropout_bp with cuda kernel helpers. * Refactored helper. * Implementation of suppresion helper with cuda kernels. * - provide cpu code fot hsvToRgb, rgbToHsv, adjustHue Signed-off-by: Yurii <yurii@skymind.io> * Using sort by value method. * Implementation of image.non_max_suppression op cuda-based helper. * - correcting and testing adjust_hue, adjust_saturation cpu/cuda code Signed-off-by: Yurii <yurii@skymind.io> * Added cuda device prefixes to declarations. * Implementation of hashcode op with cuda helper. Initital revision. * rnn cu impl removed Signed-off-by: raver119 <raver119@gmail.com>
2019-07-20 07:58:44 +02:00
RandomLauncher::fillGaussian(LaunchContext::defaultContext(), _rngA, &x0, 0.0, 1.0);
2019-06-06 14:21:15 +02:00
auto mean = x0.meanNumber(); //.e<double>(0);
auto stdev = x0.varianceNumber(nd4j::variance::SummaryStatsStandardDeviation, false);//.e<double>(0);
auto meanExp = NDArrayFactory::create<double>(0.);
auto devExp = NDArrayFactory::create<double>(1.);
ASSERT_TRUE(meanExp.equalsTo(mean, 1.e-3));
ASSERT_TRUE(devExp.equalsTo(stdev, 1.e-3));
2019-06-06 14:21:15 +02:00
}
TEST_F(RNGTests, Test_LogNormal_1) {
auto x0 = NDArrayFactory::create<float>('c', {10, 10});
auto x1 = NDArrayFactory::create<float>('c', {10, 10});
[WIP] More of CUDA operations (#69) * initial commit Signed-off-by: raver119 <raver119@gmail.com> * - gruCell_bp further Signed-off-by: Yurii <yurii@skymind.io> * - further work on gruCell_bp Signed-off-by: Yurii <yurii@skymind.io> * Inverse matrix cublas implementation. Partial working revision. * Separation of segment ops helpers. Max separation. * Separated segment_min ops. * Separation of segment_mean/sum/prod/sqrtN ops heleprs. * Fixed diagonal processing with LUP decomposition. * Modified inversion approach using current state of LU decomposition. * Implementation of matrix_inverse op with cuda kernels. Working revision. * Implemented sequence_mask cuda helper. Eliminated waste printf with matrix_inverse implementation. Added proper tests. * - further work on gruCell_bp (ff/cuda) Signed-off-by: Yurii <yurii@skymind.io> * comment one test for gruCell_bp Signed-off-by: Yurii <yurii@skymind.io> * - provide cuda static_rnn Signed-off-by: Yurii <yurii@skymind.io> * Refactored random_shuffle op to use new random generator. * Refactored random_shuffle op helper. * Fixed debug tests with random ops tests. * Implement random_shuffle op cuda kernel helper and tests. * - provide cuda scatter_update Signed-off-by: Yurii <yurii@skymind.io> * Implementation of random_shuffle for linear case with cuda kernels and tests. * Implemented random_shuffle with cuda kernels. Final revision. * - finally gruCell_bp is completed Signed-off-by: Yurii <yurii@skymind.io> * Dropout op cuda helper implementation. * Implemented dropout_bp cuda helper. * Implemented alpha_dropout_bp with cuda kernel helpers. * Refactored helper. * Implementation of suppresion helper with cuda kernels. * - provide cpu code fot hsvToRgb, rgbToHsv, adjustHue Signed-off-by: Yurii <yurii@skymind.io> * Using sort by value method. * Implementation of image.non_max_suppression op cuda-based helper. * - correcting and testing adjust_hue, adjust_saturation cpu/cuda code Signed-off-by: Yurii <yurii@skymind.io> * Added cuda device prefixes to declarations. * Implementation of hashcode op with cuda helper. Initital revision. * rnn cu impl removed Signed-off-by: raver119 <raver119@gmail.com>
2019-07-20 07:58:44 +02:00
RandomLauncher::fillLogNormal(LaunchContext::defaultContext(), _rngA, &x0, 1.0f, 2.0f);
RandomLauncher::fillLogNormal(LaunchContext::defaultContext(), _rngB, &x1, 1.0f, 2.0f);
2019-06-06 14:21:15 +02:00
ASSERT_TRUE(x0.equalsTo(&x1));
ASSERT_FALSE(x0.equalsTo(nexp0));
ASSERT_FALSE(x0.equalsTo(nexp1));
ASSERT_FALSE(x0.equalsTo(nexp2));
}
TEST_F(RNGTests, Test_Truncated_1) {
auto x0 = NDArrayFactory::create<float>('c', {10, 10});
auto x1 = NDArrayFactory::create<float>('c', {10, 10});
[WIP] More of CUDA operations (#69) * initial commit Signed-off-by: raver119 <raver119@gmail.com> * - gruCell_bp further Signed-off-by: Yurii <yurii@skymind.io> * - further work on gruCell_bp Signed-off-by: Yurii <yurii@skymind.io> * Inverse matrix cublas implementation. Partial working revision. * Separation of segment ops helpers. Max separation. * Separated segment_min ops. * Separation of segment_mean/sum/prod/sqrtN ops heleprs. * Fixed diagonal processing with LUP decomposition. * Modified inversion approach using current state of LU decomposition. * Implementation of matrix_inverse op with cuda kernels. Working revision. * Implemented sequence_mask cuda helper. Eliminated waste printf with matrix_inverse implementation. Added proper tests. * - further work on gruCell_bp (ff/cuda) Signed-off-by: Yurii <yurii@skymind.io> * comment one test for gruCell_bp Signed-off-by: Yurii <yurii@skymind.io> * - provide cuda static_rnn Signed-off-by: Yurii <yurii@skymind.io> * Refactored random_shuffle op to use new random generator. * Refactored random_shuffle op helper. * Fixed debug tests with random ops tests. * Implement random_shuffle op cuda kernel helper and tests. * - provide cuda scatter_update Signed-off-by: Yurii <yurii@skymind.io> * Implementation of random_shuffle for linear case with cuda kernels and tests. * Implemented random_shuffle with cuda kernels. Final revision. * - finally gruCell_bp is completed Signed-off-by: Yurii <yurii@skymind.io> * Dropout op cuda helper implementation. * Implemented dropout_bp cuda helper. * Implemented alpha_dropout_bp with cuda kernel helpers. * Refactored helper. * Implementation of suppresion helper with cuda kernels. * - provide cpu code fot hsvToRgb, rgbToHsv, adjustHue Signed-off-by: Yurii <yurii@skymind.io> * Using sort by value method. * Implementation of image.non_max_suppression op cuda-based helper. * - correcting and testing adjust_hue, adjust_saturation cpu/cuda code Signed-off-by: Yurii <yurii@skymind.io> * Added cuda device prefixes to declarations. * Implementation of hashcode op with cuda helper. Initital revision. * rnn cu impl removed Signed-off-by: raver119 <raver119@gmail.com>
2019-07-20 07:58:44 +02:00
RandomLauncher::fillTruncatedNormal(LaunchContext::defaultContext(), _rngA, &x0, 1.0f, 2.0f);
RandomLauncher::fillTruncatedNormal(LaunchContext::defaultContext(), _rngB, &x1, 1.0f, 2.0f);
2019-06-06 14:21:15 +02:00
ASSERT_TRUE(x0.equalsTo(&x1));
ASSERT_FALSE(x0.equalsTo(nexp0));
ASSERT_FALSE(x0.equalsTo(nexp1));
ASSERT_FALSE(x0.equalsTo(nexp2));
/* Check up distribution */
auto mean = x1.reduceNumber(reduce::Mean);
// mean.printIndexedBuffer("Mean 1.0");
auto sumA = x1 - mean; //.reduceNumber(reduce::Sum);
auto deviation = x1.varianceNumber(variance::SummaryStatsStandardDeviation, false);
//deviation /= (double)x1.lengthOf();
// deviation.printIndexedBuffer("Deviation should be 2.0");
// x1.printIndexedBuffer("Distribution TN");
}
TEST_F(RNGTests, Test_Truncated_2) {
auto x0 = NDArrayFactory::create<float>('c', {1000, 1000});
auto x1 = NDArrayFactory::create<float>('c', {1000, 1000});
[WIP] More of CUDA operations (#69) * initial commit Signed-off-by: raver119 <raver119@gmail.com> * - gruCell_bp further Signed-off-by: Yurii <yurii@skymind.io> * - further work on gruCell_bp Signed-off-by: Yurii <yurii@skymind.io> * Inverse matrix cublas implementation. Partial working revision. * Separation of segment ops helpers. Max separation. * Separated segment_min ops. * Separation of segment_mean/sum/prod/sqrtN ops heleprs. * Fixed diagonal processing with LUP decomposition. * Modified inversion approach using current state of LU decomposition. * Implementation of matrix_inverse op with cuda kernels. Working revision. * Implemented sequence_mask cuda helper. Eliminated waste printf with matrix_inverse implementation. Added proper tests. * - further work on gruCell_bp (ff/cuda) Signed-off-by: Yurii <yurii@skymind.io> * comment one test for gruCell_bp Signed-off-by: Yurii <yurii@skymind.io> * - provide cuda static_rnn Signed-off-by: Yurii <yurii@skymind.io> * Refactored random_shuffle op to use new random generator. * Refactored random_shuffle op helper. * Fixed debug tests with random ops tests. * Implement random_shuffle op cuda kernel helper and tests. * - provide cuda scatter_update Signed-off-by: Yurii <yurii@skymind.io> * Implementation of random_shuffle for linear case with cuda kernels and tests. * Implemented random_shuffle with cuda kernels. Final revision. * - finally gruCell_bp is completed Signed-off-by: Yurii <yurii@skymind.io> * Dropout op cuda helper implementation. * Implemented dropout_bp cuda helper. * Implemented alpha_dropout_bp with cuda kernel helpers. * Refactored helper. * Implementation of suppresion helper with cuda kernels. * - provide cpu code fot hsvToRgb, rgbToHsv, adjustHue Signed-off-by: Yurii <yurii@skymind.io> * Using sort by value method. * Implementation of image.non_max_suppression op cuda-based helper. * - correcting and testing adjust_hue, adjust_saturation cpu/cuda code Signed-off-by: Yurii <yurii@skymind.io> * Added cuda device prefixes to declarations. * Implementation of hashcode op with cuda helper. Initital revision. * rnn cu impl removed Signed-off-by: raver119 <raver119@gmail.com>
2019-07-20 07:58:44 +02:00
RandomLauncher::fillTruncatedNormal(LaunchContext::defaultContext(), _rngA, &x0, 1.0f, 2.0f);
RandomLauncher::fillTruncatedNormal(LaunchContext::defaultContext(), _rngB, &x1, 1.0f, 2.0f);
2019-06-06 14:21:15 +02:00
ASSERT_TRUE(x0.equalsTo(&x1));
//ASSERT_FALSE(x0.equalsTo(nexp0));
//ASSERT_FALSE(x0.equalsTo(nexp1));
//ASSERT_FALSE(x0.equalsTo(nexp2));
/* Check up distribution */
auto mean = x1.reduceNumber(reduce::Mean);
// mean.printIndexedBuffer("Mean 1.0");
//auto sumA = x1 - mean; //.reduceNumber(reduce::Sum);
auto deviation = x1.varianceNumber(variance::SummaryStatsStandardDeviation, false);
//deviation /= (double)x1.lengthOf();
// deviation.printIndexedBuffer("Deviation should be 2.0");
//x1.printIndexedBuffer("Distribution TN");
ASSERT_NEAR(mean.e<float>(0), 1.f, 0.5);
ASSERT_NEAR(deviation.e<float>(0), 2.f, 0.5);
}
TEST_F(RNGTests, Test_Truncated_21) {
auto x0 = NDArrayFactory::create<float>('c', {1000, 1000});
auto x1 = NDArrayFactory::create<float>('c', {1000, 1000});
[WIP] More of CUDA operations (#69) * initial commit Signed-off-by: raver119 <raver119@gmail.com> * - gruCell_bp further Signed-off-by: Yurii <yurii@skymind.io> * - further work on gruCell_bp Signed-off-by: Yurii <yurii@skymind.io> * Inverse matrix cublas implementation. Partial working revision. * Separation of segment ops helpers. Max separation. * Separated segment_min ops. * Separation of segment_mean/sum/prod/sqrtN ops heleprs. * Fixed diagonal processing with LUP decomposition. * Modified inversion approach using current state of LU decomposition. * Implementation of matrix_inverse op with cuda kernels. Working revision. * Implemented sequence_mask cuda helper. Eliminated waste printf with matrix_inverse implementation. Added proper tests. * - further work on gruCell_bp (ff/cuda) Signed-off-by: Yurii <yurii@skymind.io> * comment one test for gruCell_bp Signed-off-by: Yurii <yurii@skymind.io> * - provide cuda static_rnn Signed-off-by: Yurii <yurii@skymind.io> * Refactored random_shuffle op to use new random generator. * Refactored random_shuffle op helper. * Fixed debug tests with random ops tests. * Implement random_shuffle op cuda kernel helper and tests. * - provide cuda scatter_update Signed-off-by: Yurii <yurii@skymind.io> * Implementation of random_shuffle for linear case with cuda kernels and tests. * Implemented random_shuffle with cuda kernels. Final revision. * - finally gruCell_bp is completed Signed-off-by: Yurii <yurii@skymind.io> * Dropout op cuda helper implementation. * Implemented dropout_bp cuda helper. * Implemented alpha_dropout_bp with cuda kernel helpers. * Refactored helper. * Implementation of suppresion helper with cuda kernels. * - provide cpu code fot hsvToRgb, rgbToHsv, adjustHue Signed-off-by: Yurii <yurii@skymind.io> * Using sort by value method. * Implementation of image.non_max_suppression op cuda-based helper. * - correcting and testing adjust_hue, adjust_saturation cpu/cuda code Signed-off-by: Yurii <yurii@skymind.io> * Added cuda device prefixes to declarations. * Implementation of hashcode op with cuda helper. Initital revision. * rnn cu impl removed Signed-off-by: raver119 <raver119@gmail.com>
2019-07-20 07:58:44 +02:00
RandomLauncher::fillTruncatedNormal(LaunchContext::defaultContext(), _rngA, &x0, 1.0f, 2.0f);
RandomLauncher::fillTruncatedNormal(LaunchContext::defaultContext(), _rngB, &x1, 1.0f, 2.0f);
2019-06-06 14:21:15 +02:00
ASSERT_TRUE(x0.equalsTo(&x1));
auto mean0 = x0.reduceNumber(reduce::Mean);
// mean0.printIndexedBuffer("0Mean 1.0");
//auto sumA = x1 - mean; //.reduceNumber(reduce::Sum);
auto deviation0 = x0.varianceNumber(variance::SummaryStatsStandardDeviation, false);
// deviation0.printIndexedBuffer("0Deviation should be 2.0");
//ASSERT_FALSE(x0.equalsTo(nexp0));
//ASSERT_FALSE(x0.equalsTo(nexp1));
//ASSERT_FALSE(x0.equalsTo(nexp2));
/* Check up distribution */
auto mean = x1.reduceNumber(reduce::Mean);
// mean.printIndexedBuffer("Mean 1.0");
//auto sumA = x1 - mean; //.reduceNumber(reduce::Sum);
auto deviation = x1.varianceNumber(variance::SummaryStatsStandardDeviation, false);
//deviation /= (double)x1.lengthOf();
// deviation.printIndexedBuffer("Deviation should be 2.0");
//x1.printIndexedBuffer("Distribution TN");
ASSERT_NEAR(mean.e<float>(0), 1.f, 0.002);
ASSERT_NEAR(deviation.e<float>(0), 2.f, 0.5);
nd4j::ops::moments op;
auto result = op.execute({&x0}, {}, {}, {}, false, nd4j::DataType::FLOAT32);
// result->at(0)->printBuffer("MEAN");
// result->at(1)->printBuffer("VARIANCE");
delete result;
nd4j::ops::reduce_min minOp;
nd4j::ops::reduce_max maxOp;
auto minRes = minOp.execute({&x1}, {}, {}, {});
auto maxRes = maxOp.execute({&x0}, {}, {}, {});
// minRes->at(0)->printBuffer("MIN for Truncated");
// maxRes->at(0)->printBuffer("MAX for Truncated");
delete minRes;
delete maxRes;
}
TEST_F(RNGTests, Test_Truncated_22) {
auto x0 = NDArrayFactory::create<float>('c', {1000, 1000});
auto x1 = NDArrayFactory::create<float>('c', {1000, 1000});
[WIP] More of CUDA operations (#69) * initial commit Signed-off-by: raver119 <raver119@gmail.com> * - gruCell_bp further Signed-off-by: Yurii <yurii@skymind.io> * - further work on gruCell_bp Signed-off-by: Yurii <yurii@skymind.io> * Inverse matrix cublas implementation. Partial working revision. * Separation of segment ops helpers. Max separation. * Separated segment_min ops. * Separation of segment_mean/sum/prod/sqrtN ops heleprs. * Fixed diagonal processing with LUP decomposition. * Modified inversion approach using current state of LU decomposition. * Implementation of matrix_inverse op with cuda kernels. Working revision. * Implemented sequence_mask cuda helper. Eliminated waste printf with matrix_inverse implementation. Added proper tests. * - further work on gruCell_bp (ff/cuda) Signed-off-by: Yurii <yurii@skymind.io> * comment one test for gruCell_bp Signed-off-by: Yurii <yurii@skymind.io> * - provide cuda static_rnn Signed-off-by: Yurii <yurii@skymind.io> * Refactored random_shuffle op to use new random generator. * Refactored random_shuffle op helper. * Fixed debug tests with random ops tests. * Implement random_shuffle op cuda kernel helper and tests. * - provide cuda scatter_update Signed-off-by: Yurii <yurii@skymind.io> * Implementation of random_shuffle for linear case with cuda kernels and tests. * Implemented random_shuffle with cuda kernels. Final revision. * - finally gruCell_bp is completed Signed-off-by: Yurii <yurii@skymind.io> * Dropout op cuda helper implementation. * Implemented dropout_bp cuda helper. * Implemented alpha_dropout_bp with cuda kernel helpers. * Refactored helper. * Implementation of suppresion helper with cuda kernels. * - provide cpu code fot hsvToRgb, rgbToHsv, adjustHue Signed-off-by: Yurii <yurii@skymind.io> * Using sort by value method. * Implementation of image.non_max_suppression op cuda-based helper. * - correcting and testing adjust_hue, adjust_saturation cpu/cuda code Signed-off-by: Yurii <yurii@skymind.io> * Added cuda device prefixes to declarations. * Implementation of hashcode op with cuda helper. Initital revision. * rnn cu impl removed Signed-off-by: raver119 <raver119@gmail.com>
2019-07-20 07:58:44 +02:00
RandomLauncher::fillTruncatedNormal(LaunchContext::defaultContext(), _rngA, &x0, 2.0f, 4.0f);
RandomLauncher::fillTruncatedNormal(LaunchContext::defaultContext(), _rngB, &x1, 2.0f, 4.0f);
2019-06-06 14:21:15 +02:00
ASSERT_TRUE(x0.equalsTo(&x1));
auto mean0 = x0.reduceNumber(reduce::Mean);
// mean0.printIndexedBuffer("0Mean 2.0");
//auto sumA = x1 - mean; //.reduceNumber(reduce::Sum);
auto deviation0 = x0.varianceNumber(variance::SummaryStatsStandardDeviation, false);
// deviation0.printIndexedBuffer("0Deviation should be 4.0");
//ASSERT_FALSE(x0.equalsTo(nexp0));
//ASSERT_FALSE(x0.equalsTo(nexp1));
//ASSERT_FALSE(x0.equalsTo(nexp2));
/* Check up distribution */
auto mean = x1.reduceNumber(reduce::Mean);
// mean.printIndexedBuffer("Mean 2.0");
//auto sumA = x1 - mean; //.reduceNumber(reduce::Sum);
auto deviation = x1.varianceNumber(variance::SummaryStatsStandardDeviation, false);
//deviation /= (double)x1.lengthOf();
// deviation.printIndexedBuffer("Deviation should be 4.0");
//x1.printIndexedBuffer("Distribution TN");
ASSERT_NEAR(mean.e<float>(0), 2.f, 0.01);
ASSERT_NEAR(deviation.e<float>(0), 4.f, 0.52);
2019-06-06 14:21:15 +02:00
nd4j::ops::moments op;
auto result = op.execute({&x0}, {}, {}, {}, false, nd4j::DataType::FLOAT32);
// result->at(0)->printBuffer("MEAN");
// result->at(1)->printBuffer("VARIANCE");
delete result;
nd4j::ops::reduce_min minOp;
nd4j::ops::reduce_max maxOp;
auto minRes = minOp.execute({&x1}, {}, {}, {});
auto maxRes = maxOp.execute({&x0}, {}, {}, {});
// minRes->at(0)->printBuffer("MIN for Truncated2");
// maxRes->at(0)->printBuffer("MAX for Truncated2");
delete minRes;
delete maxRes;
}
TEST_F(RNGTests, Test_Truncated_23) {
auto x0 = NDArrayFactory::create<float>('c', {1000, 1000});
auto x1 = NDArrayFactory::create<float>('c', {1000, 1000});
[WIP] More of CUDA operations (#69) * initial commit Signed-off-by: raver119 <raver119@gmail.com> * - gruCell_bp further Signed-off-by: Yurii <yurii@skymind.io> * - further work on gruCell_bp Signed-off-by: Yurii <yurii@skymind.io> * Inverse matrix cublas implementation. Partial working revision. * Separation of segment ops helpers. Max separation. * Separated segment_min ops. * Separation of segment_mean/sum/prod/sqrtN ops heleprs. * Fixed diagonal processing with LUP decomposition. * Modified inversion approach using current state of LU decomposition. * Implementation of matrix_inverse op with cuda kernels. Working revision. * Implemented sequence_mask cuda helper. Eliminated waste printf with matrix_inverse implementation. Added proper tests. * - further work on gruCell_bp (ff/cuda) Signed-off-by: Yurii <yurii@skymind.io> * comment one test for gruCell_bp Signed-off-by: Yurii <yurii@skymind.io> * - provide cuda static_rnn Signed-off-by: Yurii <yurii@skymind.io> * Refactored random_shuffle op to use new random generator. * Refactored random_shuffle op helper. * Fixed debug tests with random ops tests. * Implement random_shuffle op cuda kernel helper and tests. * - provide cuda scatter_update Signed-off-by: Yurii <yurii@skymind.io> * Implementation of random_shuffle for linear case with cuda kernels and tests. * Implemented random_shuffle with cuda kernels. Final revision. * - finally gruCell_bp is completed Signed-off-by: Yurii <yurii@skymind.io> * Dropout op cuda helper implementation. * Implemented dropout_bp cuda helper. * Implemented alpha_dropout_bp with cuda kernel helpers. * Refactored helper. * Implementation of suppresion helper with cuda kernels. * - provide cpu code fot hsvToRgb, rgbToHsv, adjustHue Signed-off-by: Yurii <yurii@skymind.io> * Using sort by value method. * Implementation of image.non_max_suppression op cuda-based helper. * - correcting and testing adjust_hue, adjust_saturation cpu/cuda code Signed-off-by: Yurii <yurii@skymind.io> * Added cuda device prefixes to declarations. * Implementation of hashcode op with cuda helper. Initital revision. * rnn cu impl removed Signed-off-by: raver119 <raver119@gmail.com>
2019-07-20 07:58:44 +02:00
RandomLauncher::fillTruncatedNormal(LaunchContext::defaultContext(), _rngA, &x0, 0.0f, 1.0f);
RandomLauncher::fillTruncatedNormal(LaunchContext::defaultContext(), _rngB, &x1, 0.0f, 1.0f);
2019-06-06 14:21:15 +02:00
ASSERT_TRUE(x0.equalsTo(&x1));
auto mean0 = x0.reduceNumber(reduce::Mean);
// mean0.printIndexedBuffer("0Mean 2.0");
//auto sumA = x1 - mean; //.reduceNumber(reduce::Sum);
auto deviation0 = x0.varianceNumber(variance::SummaryStatsStandardDeviation, false);
// deviation0.printIndexedBuffer("0Deviation should be 4.0");
//ASSERT_FALSE(x0.equalsTo(nexp0));
//ASSERT_FALSE(x0.equalsTo(nexp1));
//ASSERT_FALSE(x0.equalsTo(nexp2));
/* Check up distribution */
auto mean = x1.reduceNumber(reduce::Mean);
// mean.printIndexedBuffer("Mean 2.0");
//auto sumA = x1 - mean; //.reduceNumber(reduce::Sum);
auto deviation = x1.varianceNumber(variance::SummaryStatsStandardDeviation, false);
//deviation /= (double)x1.lengthOf();
// deviation.printIndexedBuffer("Deviation should be 4.0");
//x1.printIndexedBuffer("Distribution TN");
ASSERT_NEAR(mean.e<float>(0), 0.f, 0.01);
ASSERT_NEAR(deviation.e<float>(0), 1.f, 0.5);
nd4j::ops::moments op;
auto result = op.execute({&x0}, {}, {}, {}, false, nd4j::DataType::FLOAT32);
// result->at(0)->printBuffer("MEAN");
// result->at(1)->printBuffer("VARIANCE");
delete result;
nd4j::ops::reduce_min minOp;
nd4j::ops::reduce_max maxOp;
auto minRes = minOp.execute({&x1}, {}, {}, {});
auto maxRes = maxOp.execute({&x0}, {}, {}, {});
// minRes->at(0)->printBuffer("MIN for Truncated3");
// maxRes->at(0)->printBuffer("MAX for Truncated3");
delete minRes;
delete maxRes;
}
TEST_F(RNGTests, Test_Truncated_3) {
auto x0 = NDArrayFactory::create<float>('c', {10000, 1000});
auto x1 = NDArrayFactory::create<float>('c', {10000, 1000});
[WIP] More of CUDA operations (#69) * initial commit Signed-off-by: raver119 <raver119@gmail.com> * - gruCell_bp further Signed-off-by: Yurii <yurii@skymind.io> * - further work on gruCell_bp Signed-off-by: Yurii <yurii@skymind.io> * Inverse matrix cublas implementation. Partial working revision. * Separation of segment ops helpers. Max separation. * Separated segment_min ops. * Separation of segment_mean/sum/prod/sqrtN ops heleprs. * Fixed diagonal processing with LUP decomposition. * Modified inversion approach using current state of LU decomposition. * Implementation of matrix_inverse op with cuda kernels. Working revision. * Implemented sequence_mask cuda helper. Eliminated waste printf with matrix_inverse implementation. Added proper tests. * - further work on gruCell_bp (ff/cuda) Signed-off-by: Yurii <yurii@skymind.io> * comment one test for gruCell_bp Signed-off-by: Yurii <yurii@skymind.io> * - provide cuda static_rnn Signed-off-by: Yurii <yurii@skymind.io> * Refactored random_shuffle op to use new random generator. * Refactored random_shuffle op helper. * Fixed debug tests with random ops tests. * Implement random_shuffle op cuda kernel helper and tests. * - provide cuda scatter_update Signed-off-by: Yurii <yurii@skymind.io> * Implementation of random_shuffle for linear case with cuda kernels and tests. * Implemented random_shuffle with cuda kernels. Final revision. * - finally gruCell_bp is completed Signed-off-by: Yurii <yurii@skymind.io> * Dropout op cuda helper implementation. * Implemented dropout_bp cuda helper. * Implemented alpha_dropout_bp with cuda kernel helpers. * Refactored helper. * Implementation of suppresion helper with cuda kernels. * - provide cpu code fot hsvToRgb, rgbToHsv, adjustHue Signed-off-by: Yurii <yurii@skymind.io> * Using sort by value method. * Implementation of image.non_max_suppression op cuda-based helper. * - correcting and testing adjust_hue, adjust_saturation cpu/cuda code Signed-off-by: Yurii <yurii@skymind.io> * Added cuda device prefixes to declarations. * Implementation of hashcode op with cuda helper. Initital revision. * rnn cu impl removed Signed-off-by: raver119 <raver119@gmail.com>
2019-07-20 07:58:44 +02:00
RandomLauncher::fillTruncatedNormal(LaunchContext::defaultContext(), _rngA, &x0, 1.0f, 2.0f);
RandomLauncher::fillTruncatedNormal(LaunchContext::defaultContext(), _rngB, &x1, 1.0f, 2.0f);
2019-06-06 14:21:15 +02:00
ASSERT_TRUE(x0.equalsTo(&x1));
//ASSERT_FALSE(x0.equalsTo(nexp0));
//ASSERT_FALSE(x0.equalsTo(nexp1));
//ASSERT_FALSE(x0.equalsTo(nexp2));
// Check up distribution
auto mean = x1.reduceNumber(reduce::Mean);
// mean.printIndexedBuffer("Mean 1.0");
//auto sumA = x1 - mean; //.reduceNumber(reduce::Sum);
auto deviation = x1.varianceNumber(variance::SummaryStatsStandardDeviation, false);
//deviation /= (double)x1.lengthOf();
// deviation.printIndexedBuffer("Deviation should be 2.0");
//x1.printIndexedBuffer("Distribution TN");
ASSERT_NEAR(mean.e<float>(0), 1.f, 0.001);
ASSERT_NEAR(deviation.e<float>(0), 2.f, 0.3);
}
#endif
TEST_F(RNGTests, Test_Binomial_1) {
auto x0 = NDArrayFactory::create<float>('c', {10, 10});
auto x1 = NDArrayFactory::create<float>('c', {10, 10});
RandomLauncher::fillBinomial(LaunchContext::defaultContext(), _rngA, &x0, 3, 2.0f);
RandomLauncher::fillBinomial(LaunchContext::defaultContext(), _rngB, &x1, 3, 2.0f);
2019-06-06 14:21:15 +02:00
ASSERT_TRUE(x0.equalsTo(&x1));
//nexp2->printIndexedBuffer("nexp2");
//x0.printIndexedBuffer("x0");
ASSERT_FALSE(x0.equalsTo(nexp0));
ASSERT_FALSE(x0.equalsTo(nexp1));
ASSERT_FALSE(x0.equalsTo(nexp2));
}
TEST_F(RNGTests, Test_Uniform_2) {
auto input = NDArrayFactory::create<Nd4jLong>('c', {1, 2}, {10, 10});
auto x1 = NDArrayFactory::create<float>('c', {10, 10});
RandomLauncher::fillUniform(LaunchContext::defaultContext(), _rngB, &x1, 1.0f, 2.0f);
2019-06-06 14:21:15 +02:00
auto op = new nd4j::ops::LegacyRandomOp(0);
auto result = op->execute(_rngA, {&input}, {1.0f, 2.0f}, {});
ASSERT_EQ(Status::OK(), result->status());
auto z = result->at(0);
ASSERT_TRUE(x1.isSameShape(z));
ASSERT_TRUE(x1.equalsTo(z));
delete op;
delete result;
}
TEST_F(RNGTests, Test_Gaussian_2) {
auto input = NDArrayFactory::create<Nd4jLong>('c', {1, 2}, {10, 10});
auto x1 = NDArrayFactory::create<float>('c', {10, 10});
RandomLauncher::fillGaussian(LaunchContext::defaultContext(), _rngB, &x1, 1.0f, 2.0f);
2019-06-06 14:21:15 +02:00
auto op = new nd4j::ops::LegacyRandomOp(random::GaussianDistribution);
auto result = op->execute(_rngA, {&input}, {1.0f, 2.0f}, {});
ASSERT_EQ(Status::OK(), result->status());
auto z = result->at(0);
ASSERT_TRUE(x1.isSameShape(z));
ASSERT_TRUE(x1.equalsTo(z));
delete op;
delete result;
}
TEST_F(RNGTests, Test_LogNorm_2) {
auto input = NDArrayFactory::create<Nd4jLong>('c', {1, 2}, {10, 10});
auto x1 = NDArrayFactory::create<float>('c', {10, 10});
RandomLauncher::fillLogNormal(LaunchContext::defaultContext(), _rngB, &x1, 1.0f, 2.0f);
2019-06-06 14:21:15 +02:00
auto op = new nd4j::ops::LegacyRandomOp(random::LogNormalDistribution);
auto result = op->execute(_rngA, {&input}, {1.0f, 2.0f}, {});
ASSERT_EQ(Status::OK(), result->status());
auto z = result->at(0);
ASSERT_TRUE(x1.isSameShape(z));
ASSERT_TRUE(x1.equalsTo(z));
delete op;
delete result;
}
TEST_F(RNGTests, Test_TruncatedNorm_2) {
auto input = NDArrayFactory::create<Nd4jLong>('c', {1, 2}, {10, 10});
auto x1 = NDArrayFactory::create<float>('c', {10, 10});
RandomLauncher::fillTruncatedNormal(LaunchContext::defaultContext(), _rngB, &x1, 1.0f, 2.0f);
2019-06-06 14:21:15 +02:00
auto op = new nd4j::ops::LegacyRandomOp(random::TruncatedNormalDistribution);
auto result = op->execute(_rngA, {&input}, {1.0f, 2.0f}, {});
ASSERT_EQ(Status::OK(), result->status());
auto z = result->at(0);
ASSERT_TRUE(x1.isSameShape(z));
ASSERT_TRUE(x1.equalsTo(z));
delete op;
delete result;
}
TEST_F(RNGTests, Test_Binomial_2) {
auto input = NDArrayFactory::create<Nd4jLong>('c', {1, 2}, {10, 10});
auto x1 = NDArrayFactory::create<float>('c', {10, 10});
RandomLauncher::fillBinomial(LaunchContext::defaultContext(), _rngB, &x1, 3, 0.5f);
2019-06-06 14:21:15 +02:00
auto op = new nd4j::ops::LegacyRandomOp(random::BinomialDistributionEx);
auto result = op->execute(_rngA, {&input}, {0.5f}, {3});
ASSERT_EQ(Status::OK(), result->status());
auto z = result->at(0);
ASSERT_TRUE(x1.isSameShape(z));
ASSERT_TRUE(x1.equalsTo(z));
delete op;
delete result;
}
TEST_F(RNGTests, Test_Bernoulli_2) {
auto input = NDArrayFactory::create<Nd4jLong>('c', {1, 2}, {10, 10});
auto x1 = NDArrayFactory::create<float>('c', {10, 10});
RandomLauncher::fillBernoulli(LaunchContext::defaultContext(), _rngB, &x1, 0.5f);
2019-06-06 14:21:15 +02:00
auto op = new nd4j::ops::LegacyRandomOp(random::BernoulliDistribution);
auto result = op->execute(_rngA, {&input}, {0.5f}, {});
ASSERT_EQ(Status::OK(), result->status());
auto z = result->at(0);
ASSERT_TRUE(x1.isSameShape(z));
ASSERT_TRUE(x1.equalsTo(z));
delete op;
delete result;
}
TEST_F(RNGTests, Test_GaussianDistribution_1) {
auto x = NDArrayFactory::create<Nd4jLong>('c', {2}, {10, 10});
auto exp0 = NDArrayFactory::create<float>('c', {10, 10});
nd4j::ops::random_normal op;
auto result = op.execute({&x}, {0.0, 1.0f}, {});
ASSERT_EQ(Status::OK(), result->status());
auto z = result->at(0);
ASSERT_TRUE(exp0.isSameShape(z));
ASSERT_FALSE(exp0.equalsTo(z));
ASSERT_FALSE(nexp0->equalsTo(z));
ASSERT_FALSE(nexp1->equalsTo(z));
ASSERT_FALSE(nexp2->equalsTo(z));
delete result;
}
TEST_F(RNGTests, Test_BernoulliDistribution_1) {
auto x = NDArrayFactory::create<Nd4jLong>('c', {2}, {10, 10});
auto exp0 = NDArrayFactory::create<float>('c', {10, 10});
nd4j::ops::random_bernoulli op;
auto result = op.execute({&x}, {0.5f}, {});
ASSERT_EQ(Status::OK(), result->status());
auto z = result->at(0);
ASSERT_FALSE(exp0.equalsTo(z));
ASSERT_FALSE(nexp0->equalsTo(z));
ASSERT_FALSE(nexp1->equalsTo(z));
ASSERT_FALSE(nexp2->equalsTo(z));
delete result;
}
TEST_F(RNGTests, Test_ExponentialDistribution_1) {
auto x = NDArrayFactory::create<Nd4jLong>('c', {2}, {10, 10});
auto exp0 = NDArrayFactory::create<float>('c', {10, 10});
nd4j::ops::random_exponential op;
auto result = op.execute({&x}, {0.25f}, {0});
ASSERT_EQ(Status::OK(), result->status());
auto z = result->at(0);
ASSERT_TRUE(exp0.isSameShape(z));
ASSERT_FALSE(exp0.equalsTo(z));
ASSERT_FALSE(nexp0->equalsTo(z));
ASSERT_FALSE(nexp1->equalsTo(z));
ASSERT_FALSE(nexp2->equalsTo(z));
delete result;
}
TEST_F(RNGTests, Test_ExponentialDistribution_2) {
auto x = NDArrayFactory::create<Nd4jLong>('c', {2}, {10, 10});
auto y = NDArrayFactory::create<float>('c', {10, 10});
auto exp0 = NDArrayFactory::create<float>('c', {10, 10});
y.assign(1.0);
nd4j::ops::random_exponential op;
auto result = op.execute({&x, &y}, {0.25f}, {0});
ASSERT_EQ(Status::OK(), result->status());
auto z = result->at(0);
ASSERT_TRUE(exp0.isSameShape(z));
ASSERT_FALSE(exp0.equalsTo(z));
ASSERT_FALSE(nexp0->equalsTo(z));
ASSERT_FALSE(nexp1->equalsTo(z));
ASSERT_FALSE(nexp2->equalsTo(z));
delete result;
}
TEST_F(RNGTests, Test_PoissonDistribution_1) {
auto x = NDArrayFactory::create<Nd4jLong>('c', {1}, {10});
auto la = NDArrayFactory::create<float>('c', {2, 3});
auto exp0 = NDArrayFactory::create<float>('c', {10, 2, 3});
la.linspace(1.0);
nd4j::ops::random_poisson op;
auto result = op.execute({&x, &la}, {}, {});
ASSERT_EQ(Status::OK(), result->status());
auto z = result->at(0);
// z->printIndexedBuffer("Poisson distribution");
ASSERT_TRUE(exp0.isSameShape(z));
ASSERT_FALSE(exp0.equalsTo(z));
delete result;
}
TEST_F(RNGTests, Test_GammaDistribution_1) {
auto x = NDArrayFactory::create<Nd4jLong>('c', {1}, {10});
auto al = NDArrayFactory::create<float>('c', {2, 3});
auto exp0 = NDArrayFactory::create<float>('c', {10, 2, 3});
al.linspace(1.0);
nd4j::ops::random_gamma op;
auto result = op.execute({&x, &al}, {}, {});
ASSERT_EQ(Status::OK(), result->status());
auto z = result->at(0);
// z->printIndexedBuffer("Gamma distribution");
ASSERT_TRUE(exp0.isSameShape(z));
ASSERT_FALSE(exp0.equalsTo(z));
delete result;
}
TEST_F(RNGTests, Test_GammaDistribution_2) {
auto x = NDArrayFactory::create<Nd4jLong>('c', {1}, {10});
auto al = NDArrayFactory::create<float>('c', {2, 3});
auto be = NDArrayFactory::create<float>('c', {2, 3});
auto exp0 = NDArrayFactory::create<float>('c', {10, 2, 3});
al.linspace(1.0);
be.assign(1.0);
nd4j::ops::random_gamma op;
auto result = op.execute({&x, &al, &be}, {}, {});
ASSERT_EQ(Status::OK(), result->status());
auto z = result->at(0);
// z->printIndexedBuffer("Gamma distribution");
ASSERT_TRUE(exp0.isSameShape(z));
ASSERT_FALSE(exp0.equalsTo(z));
delete result;
}
TEST_F(RNGTests, Test_GammaDistribution_3) {
auto x = NDArrayFactory::create<Nd4jLong>('c', {1}, {10});
auto al = NDArrayFactory::create<float>('c', {3, 1});
auto be = NDArrayFactory::create<float>('c', {1, 2});
auto exp0 = NDArrayFactory::create<float>('c', {10, 3, 2});
al.linspace(1.0);
be.assign(2.0);
nd4j::ops::random_gamma op;
auto result = op.execute({&x, &al, &be}, {}, {});
ASSERT_EQ(Status::OK(), result->status());
auto z = result->at(0);
// z->printIndexedBuffer("Gamma distribution");
ASSERT_TRUE(exp0.isSameShape(z));
ASSERT_FALSE(exp0.equalsTo(z));
2019-06-06 14:21:15 +02:00
delete result;
}
TEST_F(RNGTests, Test_UniformDistribution_04) {
auto x = NDArrayFactory::create<Nd4jLong>('c', {1}, {10});
auto al = NDArrayFactory::create<int>(1);
auto be = NDArrayFactory::create<int>(20);
auto exp0 = NDArrayFactory::create<float>('c', {10});
nd4j::ops::randomuniform op;
auto result = op.execute({&x, &al, &be}, {}, {DataType::INT32});
ASSERT_EQ(Status::OK(), result->status());
auto z = result->at(0);
z->printIndexedBuffer("Uniform int distribution");
ASSERT_TRUE(exp0.isSameShape(z));
ASSERT_FALSE(exp0.equalsTo(z));
delete result;
}
2019-06-06 14:21:15 +02:00
namespace nd4j {
namespace tests {
static void fillList(Nd4jLong seed, int numberOfArrays, std::vector<Nd4jLong> &shape, std::vector<NDArray*> &list, nd4j::graph::RandomGenerator *rng) {
rng->setSeed((int) seed);
[WIP] More of CUDA (#95) * initial commit Signed-off-by: raver119 <raver119@gmail.com> * Implementation of hashcode cuda helper. Working edition. * Fixed parallel test input arangements. * Fixed tests for hashcode op. * Fixed shape calculation for image:crop_and_resize op and test. * NativeOps tests. Initial test suite. * Added tests for indexReduce methods. * Added test on execBroadcast with NDArray as dimensions. * Added test on execBroadcastBool with NDArray as dimensions. * Added tests on execPairwiseTransform and execPairwiseTransofrmBool. * Added tests for execReduce with scalar results. * Added reduce tests for non-empty dims array. * Added tests for reduce3. * Added tests for execScalar. * Added tests for execSummaryStats. * - provide cpu/cuda code for batch_to_space - testing it Signed-off-by: Yurii <yurii@skymind.io> * - remove old test for batch_to_space (had wrong format and numbers were not checked) Signed-off-by: Yurii <yurii@skymind.io> * Fixed complilation errors with test. * Added test for execTransformFloat. * Added test for execTransformSame. * Added test for execTransformBool. * Added test for execTransformStrict. * Added tests for execScalar/execScalarBool with TADs. * Added test for flatten. * - provide cpu/cuda code for space_to_Batch operaion Signed-off-by: Yurii <yurii@skymind.io> * Added test for concat. * comment unnecessary stuff in s_t_b Signed-off-by: Yurii <yurii@skymind.io> * Added test for specialConcat. * Added tests for memcpy/set routines. * Fixed pullRow cuda test. * Added pullRow test. * Added average test. * - correct typo in NDArray::applyPairwiseTransform(nd4j::pairwise::BoolOps op...) Signed-off-by: Yurii <yurii@skymind.io> * - debugging and fixing cuda tests in JavaInteropTests file Signed-off-by: Yurii <yurii@skymind.io> * - correct some tests Signed-off-by: Yurii <yurii@skymind.io> * Added test for shuffle. * Fixed ops declarations. * Restored omp and added shuffle test. * Added convertTypes test. * Added tests for execRandom. Eliminated usage of RandomBuffer with NativeOps. * Added sort tests. * Added tests for execCustomOp. * - further debuging and fixing tests terminated with crash Signed-off-by: Yurii <yurii@skymind.io> * Added tests for calculateOutputShapes. * Addded Benchmarks test. * Commented benchmark tests. * change assertion Signed-off-by: raver119 <raver119@gmail.com> * Added tests for apply_sgd op. Added cpu helper for that op. * Implement cuda helper for aplly_sgd op. Fixed tests for NativeOps. * Added test for assign broadcastable. * Added tests for assign_bp op. * Added tests for axpy op. * - assign/execScalar/execTransformAny signature change - minor test fix Signed-off-by: raver119 <raver119@gmail.com> * Fixed axpy op. * meh Signed-off-by: raver119 <raver119@gmail.com> * - fix tests for nativeOps::concat Signed-off-by: Yurii <yurii@skymind.io> * sequential transform/scalar Signed-off-by: raver119 <raver119@gmail.com> * allow nested parallelism Signed-off-by: raver119 <raver119@gmail.com> * assign_bp leak fix Signed-off-by: raver119 <raver119@gmail.com> * block setRNG fix Signed-off-by: raver119 <raver119@gmail.com> * enable parallelism by default Signed-off-by: raver119 <raver119@gmail.com> * enable nested parallelism by default Signed-off-by: raver119 <raver119@gmail.com> * Added cuda implementation for row_count helper. * Added implementation for tnse gains op helper. * - take into account possible situations when input arrays are empty in reduce_ cuda stuff Signed-off-by: Yurii <yurii@skymind.io> * Implemented tsne/edge_forces op cuda-based helper. Parallelized cpu-based helper for edge_forces. * Added kernel for tsne/symmetrized op heleper. * Implementation of tsne/symmetrized op cuda helper. Working edition. * Eliminated waste printfs. * Added test for broadcastgradientargs op. * host-only fallback for empty reduce float Signed-off-by: raver119 <raver119@gmail.com> * - some tests fixes Signed-off-by: Yurii <yurii@skymind.io> * - correct the rest of reduce_ stuff Signed-off-by: Yurii <yurii@skymind.io> * - further correction of reduce_ stuff Signed-off-by: Yurii <yurii@skymind.io> * Added test for Cbow op. Also added cuda implementation for cbow helpers. * - improve code of stack operation for scalar case Signed-off-by: Yurii <yurii@skymind.io> * - provide cuda kernel for gatherND operation Signed-off-by: Yurii <yurii@skymind.io> * Implementation of cbow helpers with cuda kernels. * minor tests tweaks Signed-off-by: raver119 <raver119@gmail.com> * minor tests tweaks Signed-off-by: raver119 <raver119@gmail.com> * - further correction of cuda stuff Signed-off-by: Yurii <yurii@skymind.io> * Implementatation of cbow op helper with cuda kernels. Working edition. * Skip random testing for cudablas case. * lstmBlockCell context fix Signed-off-by: raver119 <raver119@gmail.com> * Added tests for ELU and ELU_BP ops. * Added tests for eq_scalar, gt_scalar, gte_scalar and lte_scalar ops. * Added tests for neq_scalar. * Added test for noop. * - further work on clipbynorm_bp Signed-off-by: Yurii <yurii@skymind.io> * - get rid of concat op call, use instead direct concat helper call Signed-off-by: Yurii <yurii@skymind.io> * lstmBlockCell context fix Signed-off-by: raver119 <raver119@gmail.com> * Added tests for lrelu and lrelu_bp. * Added tests for selu and selu_bp. * Fixed lrelu derivative helpers. * - some corrections in lstm Signed-off-by: Yurii <yurii@skymind.io> * operator * result shape fix Signed-off-by: raver119 <raver119@gmail.com> * - correct typo in lstmCell Signed-off-by: Yurii <yurii@skymind.io> * few tests fixed Signed-off-by: raver119 <raver119@gmail.com> * CUDA inverse broadcast bool fix Signed-off-by: raver119 <raver119@gmail.com> * disable MMAP test for CUDA Signed-off-by: raver119 <raver119@gmail.com> * BooleanOp syncToDevice Signed-off-by: raver119 <raver119@gmail.com> * meh Signed-off-by: raver119 <raver119@gmail.com> * additional data types for im2col/col2im Signed-off-by: raver119 <raver119@gmail.com> * Added test for firas_sparse op. * one more RandomBuffer test excluded Signed-off-by: raver119 <raver119@gmail.com> * Added tests for flatten op. * Added test for Floor op. * bunch of tests fixed Signed-off-by: raver119 <raver119@gmail.com> * mmulDot tests fixed Signed-off-by: raver119 <raver119@gmail.com> * more tests fixed Signed-off-by: raver119 <raver119@gmail.com> * Implemented floordiv_bp op and tests. * Fixed scalar case with cuda implementation for bds. * - work on cuda kernel for clip_by_norm backprop op is completed Signed-off-by: Yurii <yurii@skymind.io> * Eliminate cbow crach. * more tests fixed Signed-off-by: raver119 <raver119@gmail.com> * more tests fixed Signed-off-by: raver119 <raver119@gmail.com> * Eliminated abortion with batched nlp test. * more tests fixed Signed-off-by: raver119 <raver119@gmail.com> * Fixed shared flag initializing. * disabled bunch of cpu workspaces tests Signed-off-by: raver119 <raver119@gmail.com> * scalar operators fix: missing registerSpecialUse call Signed-off-by: raver119 <raver119@gmail.com> * Fixed logdet for cuda and tests. * - correct clipBynorm_bp Signed-off-by: Yurii <yurii@skymind.io> * Fixed crop_and_resize shape datatype. * - correct some mmul tests Signed-off-by: Yurii <yurii@skymind.io>
2019-08-02 19:01:03 +02:00
2019-06-06 14:21:15 +02:00
for (int i = 0; i < numberOfArrays; i++) {
auto arrayI = NDArrayFactory::create<Nd4jLong>(shape);
auto arrayR = NDArrayFactory::create_<double>('c', shape);
auto min = NDArrayFactory::create(0.0);
auto max = NDArrayFactory::create(1.0);
2019-06-06 14:21:15 +02:00
nd4j::ops::randomuniform op;
op.execute(*rng, {&arrayI, &min, &max}, {arrayR}, {}, {DataType::DOUBLE}, {}, false);
2019-06-06 14:21:15 +02:00
list.emplace_back(arrayR);
2019-06-06 14:21:15 +02:00
}
};
}
}
TEST_F(RNGTests, Test_Reproducibility_1) {
Nd4jLong seed = 123;
std::vector<Nd4jLong> shape = {32, 3, 28, 28};
nd4j::graph::RandomGenerator rng;
std::vector<NDArray*> expList;
nd4j::tests::fillList(seed, 10, shape, expList, &rng);
for (int e = 0; e < 2; e++) {
std::vector<NDArray *> trialList;
nd4j::tests::fillList(seed, 10, shape, trialList, &rng);
for (int a = 0; a < expList.size(); a++) {
auto arrayE = expList[a];
auto arrayT = trialList[a];
bool t = arrayE->equalsTo(arrayT);
if (!t) {
// nd4j_printf("Failed at iteration [%i] for array [%i]\n", e, a);
ASSERT_TRUE(false);
}
delete arrayT;
}
}
for (auto v: expList)
delete v;
}
#ifndef DEBUG_BUILD
TEST_F(RNGTests, Test_Reproducibility_2) {
Nd4jLong seed = 123;
std::vector<Nd4jLong> shape = {32, 3, 64, 64};
nd4j::graph::RandomGenerator rng;
std::vector<NDArray*> expList;
nd4j::tests::fillList(seed, 10, shape, expList, &rng);
for (int e = 0; e < 2; e++) {
std::vector<NDArray*> trialList;
nd4j::tests::fillList(seed, 10, shape, trialList, &rng);
for (int a = 0; a < expList.size(); a++) {
auto arrayE = expList[a];
auto arrayT = trialList[a];
bool t = arrayE->equalsTo(arrayT);
if (!t) {
// nd4j_printf("Failed at iteration [%i] for array [%i]\n", e, a);
for (Nd4jLong f = 0; f < arrayE->lengthOf(); f++) {
double x = arrayE->e<double>(f);
double y = arrayT->e<double>(f);
if (nd4j::math::nd4j_re(x, y) > 0.1) {
// nd4j_printf("E[%lld] %f != T[%lld] %f\n", (long long) f, (float) x, (long long) f, (float) y);
throw std::runtime_error("boom");
}
}
// just breaker, since test failed
ASSERT_TRUE(false);
}
delete arrayT;
}
}
for (auto v: expList)
delete v;
}
TEST_F(RNGTests, Test_Uniform_4) {
auto x1 = NDArrayFactory::create<double>('c', {1000000});
[WIP] More of CUDA operations (#69) * initial commit Signed-off-by: raver119 <raver119@gmail.com> * - gruCell_bp further Signed-off-by: Yurii <yurii@skymind.io> * - further work on gruCell_bp Signed-off-by: Yurii <yurii@skymind.io> * Inverse matrix cublas implementation. Partial working revision. * Separation of segment ops helpers. Max separation. * Separated segment_min ops. * Separation of segment_mean/sum/prod/sqrtN ops heleprs. * Fixed diagonal processing with LUP decomposition. * Modified inversion approach using current state of LU decomposition. * Implementation of matrix_inverse op with cuda kernels. Working revision. * Implemented sequence_mask cuda helper. Eliminated waste printf with matrix_inverse implementation. Added proper tests. * - further work on gruCell_bp (ff/cuda) Signed-off-by: Yurii <yurii@skymind.io> * comment one test for gruCell_bp Signed-off-by: Yurii <yurii@skymind.io> * - provide cuda static_rnn Signed-off-by: Yurii <yurii@skymind.io> * Refactored random_shuffle op to use new random generator. * Refactored random_shuffle op helper. * Fixed debug tests with random ops tests. * Implement random_shuffle op cuda kernel helper and tests. * - provide cuda scatter_update Signed-off-by: Yurii <yurii@skymind.io> * Implementation of random_shuffle for linear case with cuda kernels and tests. * Implemented random_shuffle with cuda kernels. Final revision. * - finally gruCell_bp is completed Signed-off-by: Yurii <yurii@skymind.io> * Dropout op cuda helper implementation. * Implemented dropout_bp cuda helper. * Implemented alpha_dropout_bp with cuda kernel helpers. * Refactored helper. * Implementation of suppresion helper with cuda kernels. * - provide cpu code fot hsvToRgb, rgbToHsv, adjustHue Signed-off-by: Yurii <yurii@skymind.io> * Using sort by value method. * Implementation of image.non_max_suppression op cuda-based helper. * - correcting and testing adjust_hue, adjust_saturation cpu/cuda code Signed-off-by: Yurii <yurii@skymind.io> * Added cuda device prefixes to declarations. * Implementation of hashcode op with cuda helper. Initital revision. * rnn cu impl removed Signed-off-by: raver119 <raver119@gmail.com>
2019-07-20 07:58:44 +02:00
RandomLauncher::fillUniform(LaunchContext::defaultContext(), _rngB, &x1, 1.0, 2.0);
2019-06-06 14:21:15 +02:00
/* Check up distribution */
auto mean = x1.reduceNumber(reduce::Mean);
// mean.printIndexedBuffer("Mean should be 1.5");
auto sumA = x1 - mean; //.reduceNumber(reduce::Sum);
auto deviation = x1.varianceNumber(variance::SummaryStatsVariance, false);
//deviation /= (double)x1.lengthOf();
// deviation.printIndexedBuffer("Deviation should be 1/12 (0.083333)");
ASSERT_NEAR(mean.e<double>(0), 1.5, 1e-3);
ASSERT_NEAR(1/12., deviation.e<double>(0), 1e-3);
}
#endif
TEST_F(RNGTests, test_choice_1) {
auto x = NDArrayFactory::linspace<double>(0, 10, 11);
auto prob = NDArrayFactory::valueOf<double>({11}, 1.0/11, 'c');
auto z = NDArrayFactory::create<double>('c', {1000});
RandomGenerator rng(119, 256);
NativeOpExecutioner::execRandom(nd4j::LaunchContext ::defaultContext(), random::Choice, &rng, x->buffer(), x->shapeInfo(), x->specialBuffer(), x->specialShapeInfo(), prob->buffer(), prob->shapeInfo(), prob->specialBuffer(), prob->specialShapeInfo(), z.buffer(), z.shapeInfo(), z.specialBuffer(), z.specialShapeInfo(), nullptr);
// z.printIndexedBuffer("z");
delete x;
delete prob;
}