cavis/libnd4j/blas/cpu/NDArrayFactory.cpp

621 lines
36 KiB
C++
Raw Normal View History

2019-06-06 14:21:15 +02:00
/*******************************************************************************
* Copyright (c) 2015-2018 Skymind, Inc.
*
* 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
******************************************************************************/
//
// Created by GS <sgazeos@gmail.com> on 2018-12-20.
//
#include <NDArrayFactory.h>
#include <exceptions/cuda_exception.h>
#include <ConstantHelper.h>
#include <ConstantShapeHelper.h>
#include <ShapeUtils.h>
#include <type_traits>
namespace nd4j {
Merge master to upstream (#7945) * Shugeo strided slice zeros (#14) * Modified strided_slice op to properly work with empty-like shapes. * Fixed test for reduce_mean with empty-like input. * [WIP] Last merge (#15) * correct logsoftmax looss (#2) * Small SameDiff listener fix (#4) * Various fixes (#6) * #7839 Fix for asXMatrix and tests * #7866 EmbeddingSequenceLayer dtype fix + test * #7856 SameDiff save/load stream methods * #7859 RegressionEvaluation rank 4 fix + tests + axis configuration * EvaluationBinary 3d/4d * More evaluation 3d/4d tests * #7847 Evaluation empty checks * Small test ifx * #7848 Fix median edge case * Improve DL4J samediff layer tests * [WIP] FastText wrapper implemented (#8) * FastText implemented * Some fixes * Fix shapes for wordsNearest * Validation of input vectors * Fixes * Fixed test * Thread tagged * Some tweaks * setContextClassLoader for DeallocatorServiceThread * Numpy format tests (#1) * Various fixes (#11) * #7852 SameDiff gather fix * #7892 SameDiff placeholder to constant conversion * #7890 validate input rank for MLN/CG init methods * Fix broken permute shape calculation * Permute and gather fixes * Tests * #7850 LogSumExp fix + test * Handful of test fixes * Empty arrays with non-scalar shapes (#10) * minor rearrangements for lambdas * empty tensors with non-scalar shapes * numpy empty tensors with non-scalar shapes * few more empty tweaks * Small fixes * conv3d signature update * micro fix in batchnorm mkldnn * Import fixes * Fix * MKL-DNN update * Small fill fix * fill with empty input + test * Fixes * Small error improvement * Fix * one special test * couple of fixes for lstm * Rewrite TFGraphMapper.getNDArrayFromTensor to be maintainable and less error prone * Fixes * FP16 * Unsigned * BFloat16 * Fill op - empty tweaks * - couple of fixes for empty arrays construction - stack updated * strided slice fix * one transform test * provide method for reducing shapeInfo in case of input array is empty * Fixed reduceAlongDimensions to use empty input properly. * couple of broadcast tests * couple of tests broadcast tests + tweak to make them pass * add check of non-empty to methods producing sub-arrays * Fixed reshapeC with zeros in shape. * complete empty check in reduce_... legacy ops * Concat and cumsum/prod * Tweak to empty shape inference on import * add empty check to the rest of reduce legacy ops * one more test * correct typo in evalReduceShapeInfoEmpty * Added tests for reduce_* ops to tests with zero shapes. * few more tests for empty reductions * Fixed strided_slice op with empty case and tests. * one more empty reduction test * Fixed strided_slice test. * add empty check to NDArray::reshapei * infOrMax * empty min/max with infinity tests * made unstack working correctly with empty arrays * few IndexReduce tests + tweaks for empty shapes * add test for empty concat * few tests fixed * Validation fix for reductions on empty shapes * Reverse fix * Reduction shape calc fixes * SameDiff.generateOutputVariable: don't use shape function to determine number of outputs * Range fix * - NDArray constructor updated for scalars/empty arrays - few tests fixed * More fixes * Empty creator fixes * concat fix * concat fix * TF import tests: allow 'both all NaN' and 'both all inf' to pass * Slice, zero fraction, and reshape fixes * transpose, gather * Zero fraction * scalar cast fix * Empty reduction axis support * few more tests fixed * Fixed input checks conforming with TF for concat op and tests. * few tests fixed * matmul scalar shape fix * Fixed checkout for data type and scalarity with concat to allow non-empty scalars with vector concats. * broadcast bool fix * few more tests * few more tests * correct evalReduceShapeInfoEmpty * argmax/argmin + tests * one more empty edge case + one more test * argmax/argmin/realdiv_bp tweaks * empty reshape test + fix * Helper fixes * Small fixes * Gather test fix * Gather test fix * Small fixes * reduce scalar zero values * scalar mean workaround * Remove debug code * along dim mean workaround * one more test * - equalsTo() tweak for empty arrays - one more test * broadcast tweaks * [WIP] Fixing outstanding issues for NLP (#9) * Avoid using not-inited objects * Test fixed. * Redundant method avoided for models like FastText * KMeans++ implementation * KMeans++ implementation * Disable parallel execution * KMeans++ * Tests * Dev branch merge (#16) * SameDiff: convertDataType and gradient check util improvements (#12) * GradCheck util improvements * StopGradient constructor + test * SameDiff: Add datatype conversion * Javadoc and add DataType.isNumerical() * Small fix * Fix SameDiff TF import test cases intermediate naming (workaround for bad default) * TFGraphTestAllHelper: check intermediates in execution order * Add missing debug listener * [WIP] lstmBlock fix + other changes (#13) - fixes lstmBlock issue - changes NDArray method reshape(), permute(), transpose() by making them return instance instead of pointer - CheckNumerics op - fixes for ReduceBool IsInfOrNan & IsFinite * Small test fix * CheckNumerics op wrapper * Fix some issues on master (#17) * Fix DataVec test issue * Fix issue with dl4j SameDiff output layer * Dtype fix for lambda layers * #7912 BertIterator dtype fix (use float32 not global default) * [WIP] Next set of CUDA stuff (#7) New CUDA implementations and improvements * bad file * Dev branch master merge (#23) * SameDiff: convertDataType and gradient check util improvements (#12) * GradCheck util improvements * StopGradient constructor + test * SameDiff: Add datatype conversion * Javadoc and add DataType.isNumerical() * Small fix * Fix SameDiff TF import test cases intermediate naming (workaround for bad default) * TFGraphTestAllHelper: check intermediates in execution order * Add missing debug listener * [WIP] lstmBlock fix + other changes (#13) - fixes lstmBlock issue - changes NDArray method reshape(), permute(), transpose() by making them return instance instead of pointer - CheckNumerics op - fixes for ReduceBool IsInfOrNan & IsFinite * Small test fix * CheckNumerics op wrapper * Compatibility of deserialization (#18) Signed-off-by: Alexander Stoyakin <alexander.stoyakin@gmail.com> * SameDiff: add activation gradient checking support for debugging (#19) * SameDiff gradient checker: first pass on activation gradient checks * Fixes + tests for activation gradient checking * Javadoc * [WIP] Some nd4j data type corrections (#20) * Adjust data type * Set correct Data type. * Size of proper data type. * fix averaged cpu load (#22) * SameDiff ops, TF import and fixes (#24) * CheckNumerics tests + fixes + misc fixes Signed-off-by: AlexDBlack <blacka101@gmail.com> * Fake quant Signed-off-by: AlexDBlack <blacka101@gmail.com> * Fixes Signed-off-by: AlexDBlack <blacka101@gmail.com> * FakeQuantWithMinMaxArgs Signed-off-by: AlexDBlack <blacka101@gmail.com> * CheckNumerics fix Signed-off-by: AlexDBlack <blacka101@gmail.com> * Fix libnd4j ALL_INTS and ALL_FLOATS declaration (uint and bfloat types) Signed-off-by: AlexDBlack <blacka101@gmail.com> * Small fix Signed-off-by: AlexDBlack <blacka101@gmail.com> * Javadoc Signed-off-by: AlexDBlack <blacka101@gmail.com> * Exception tweak Signed-off-by: AlexDBlack <blacka101@gmail.com> * fix Signed-off-by: AlexDBlack <blacka101@gmail.com> * Fix for out of scope stack allocated var use Signed-off-by: AlexDBlack <blacka101@gmail.com> * Ignores Signed-off-by: AlexDBlack <blacka101@gmail.com> * Ignore for known failing test (already logged issue) Signed-off-by: AlexDBlack <blacka101@gmail.com> * Merge upstream to fork (#25) * Add thousand-separator commas to TotalParams (#7915) * Add thousand-separator commas to TotalParams The number of parameters can be quite large, and it would help the reading of the summary printout to have the TotalParams column & values at the bottom have thousand-separator-commas in them. * Add thousand-separator commas to MultiLayerNetwork Corresponding change to MultiLayerNetwork Signed-off-by: Jxtps Jxtps <jxtps435@gmail.com> * Update contributing and issue/PR templates (#7934) Signed-off-by: AlexDBlack <blacka101@gmail.com> * Fix link to AdaDelta paper (#7942) Fix link to AdaDelta paper hosted on matthewzeiler.com Signed-off-by: Jxtps * Fixes, and ignores for known/logged failing issues (#7943) Signed-off-by: AlexDBlack <blacka101@gmail.com> * SameDiff + DL4J/SameDiff: Multiple fixes (#28) * #7919 HDF5 attribute buffer length fix Signed-off-by: AlexDBlack <blacka101@gmail.com> * #7909 Arbiter constructor exception ux improvements Signed-off-by: AlexDBlack <blacka101@gmail.com> * #7925 RNN output layer length checks Signed-off-by: AlexDBlack <blacka101@gmail.com> * #7939 Add listener for validating inputs are not incorrectly modified Signed-off-by: AlexDBlack <blacka101@gmail.com> * #7939 Integrate NonInplaceValidationListener into tests * #7844 DL4J SameDiff fixes for variable minibatch size * DL4J SameDiff fixes - ensure gradient for input placeholder is available Signed-off-by: AlexDBlack <blacka101@gmail.com> * Tweaks to ExternalErrorsFunction - use placeholders, make more robust * Another fix * More fixes * More SameDiff/DL4J fixes * Scope out scalar array creation in BaseScalarOp * Remove debug code Signed-off-by: AlexDBlack <blacka101@gmail.com> * [WIP] Final dev branch merge (#29) * SameDiff: convertDataType and gradient check util improvements (#12) * GradCheck util improvements * StopGradient constructor + test * SameDiff: Add datatype conversion * Javadoc and add DataType.isNumerical() * Small fix * Fix SameDiff TF import test cases intermediate naming (workaround for bad default) * TFGraphTestAllHelper: check intermediates in execution order * Add missing debug listener * [WIP] lstmBlock fix + other changes (#13) - fixes lstmBlock issue - changes NDArray method reshape(), permute(), transpose() by making them return instance instead of pointer - CheckNumerics op - fixes for ReduceBool IsInfOrNan & IsFinite * Small test fix * CheckNumerics op wrapper * Compatibility of deserialization (#18) Signed-off-by: Alexander Stoyakin <alexander.stoyakin@gmail.com> * SameDiff: add activation gradient checking support for debugging (#19) * SameDiff gradient checker: first pass on activation gradient checks * Fixes + tests for activation gradient checking * Javadoc * [WIP] Some nd4j data type corrections (#20) * Adjust data type * Set correct Data type. * Size of proper data type. * fix averaged cpu load (#22) * [WIP] Multiple dataset iterators (#27) * Splitting dataset into arbitrary number * Fixes * Multiple split of iterator * Test * Test * Some fixes * signature change * one more tweak Signed-off-by: raver119 <raver119@gmail.com> * one more test for sequential use of DataSetIteratorSplitter Signed-off-by: raver119 <raver119@gmail.com> * Fixes * Fixes * one more test for Alexander Signed-off-by: raver119 <raver119@gmail.com> * Some fixes * Some fixes * one more test for Alexander Signed-off-by: raver119 <raver119@gmail.com> * minor test fix Signed-off-by: raver119 <raver119@gmail.com> * Some fixes * Some fixes * couple of assertions tweaked Signed-off-by: raver119 <raver119@gmail.com> * MDS splitter test :/ Signed-off-by: raver119 <raver119@gmail.com> * Minor refactoring * Multi dataset * Some fixes * More tests * Small number of test fixes/improvements (failures on CI) (#31) Signed-off-by: AlexDBlack <blacka101@gmail.com> * [WIP] More CUDA stuff (#26) * initial commit Signed-off-by: raver119 <raver119@gmail.com> * LRN BP CUDA Signed-off-by: raver119 <raver119@gmail.com> * less memory Signed-off-by: raver119 <raver119@gmail.com> * Fixed bug with crop_and_resize op helper. * get rid of unnecessary index-calculation dunction Signed-off-by: Yurii <yurii@skymind.io> * Fixed sort with nth_element cuda-based helper. * Refactored nth_element. * Refactored nth_element op and tests. * Modified usage of dim array with sortTad routine. * Refactored main routine of helper for non_max_image_suppression op. * non_max_image_suppression op helper with cuda kernel implementation. Initial revision. * fix vol2col cuda kernel * meh Signed-off-by: raver119 <raver119@gmail.com> * topK concept Signed-off-by: raver119 <raver119@gmail.com> * unsorted topK with scanWitdh of 1 Signed-off-by: raver119 <raver119@gmail.com> * correct vol2col tests * sorted/unsorted topK Signed-off-by: raver119 <raver119@gmail.com> * implementation and fixing col2im/col2vol * Corrected usage flags with input/output with reverse op. * dup is const now Signed-off-by: raver119 <raver119@gmail.com> * percentile op Signed-off-by: raver119 <raver119@gmail.com> * group tests for mapool2d Signed-off-by: Yurii <yurii@skymind.io> * special test for george Signed-off-by: raver119 <raver119@gmail.com> * less threads for sortTad Signed-off-by: raver119 <raver119@gmail.com> * provide conv2d for cuda Signed-off-by: Yurii <yurii@skymind.io> * remove auther in sort tad kernel code Signed-off-by: Yurii <yurii@skymind.io> * provide depthwise_conv2d for cuda Signed-off-by: Yurii <yurii@skymind.io> * - max_pooling_with_argmax - null check for special use Signed-off-by: raver119 <raver119@gmail.com> * dts cuda Signed-off-by: raver119 <raver119@gmail.com> * provide sconv2d for cuda Signed-off-by: Yurii <yurii@skymind.io> * std cuda Signed-off-by: raver119 <raver119@gmail.com> * Refactored non_max_suppression op to conform TF implementation. * Improved suppression helper. * provide pooling3d for cuda Signed-off-by: Yurii <yurii@skymind.io> * minor lstm rearrangements Signed-off-by: raver119 <raver119@gmail.com> * more of minor lstm rearrangements Signed-off-by: raver119 <raver119@gmail.com> * (bi)dynamic_rnn Signed-off-by: raver119 <raver119@gmail.com> * templates init order Signed-off-by: raver119 <raver119@gmail.com> * Refactored non_max_suppression op. * Added cuda kernel for non_max_suppression. * CPU sort by key/value Signed-off-by: raver119 <raver119@gmail.com> * CPU sort TAD by key/value Signed-off-by: raver119 <raver119@gmail.com> * CPU sort TAD by key/value tests Signed-off-by: raver119 <raver119@gmail.com> * Eliminate compiler error with cuda implementation. * - repaired gradCheck in cuda - provide conv2d_bp for cuda Signed-off-by: Yurii <yurii@skymind.io> * missed signature Signed-off-by: raver119 <raver119@gmail.com> * provide depthwise_conv2d_bp for cuda Signed-off-by: Yurii <yurii@skymind.io> * Implementation of lup helper with cuda kernel. Initial commit. * further work on backprops for convolutions Signed-off-by: Yurii <yurii@skymind.io> * CUDA linear sort by key/val Signed-off-by: raver119 <raver119@gmail.com> * CUDA tad sort by key/val Signed-off-by: raver119 <raver119@gmail.com> * start providing of backprop for pooling2d/3d Signed-off-by: Yurii <yurii@skymind.io> * Added atomicAdd for bool datatype. * dynamic partition concept Signed-off-by: raver119 <raver119@gmail.com> * dynamic partition concept Signed-off-by: raver119 <raver119@gmail.com> * dynamic partition scalar CUDA Signed-off-by: raver119 <raver119@gmail.com> * important comment Signed-off-by: raver119 <raver119@gmail.com> * fix pooling2d/3d backprop helpers Signed-off-by: Yurii <yurii@skymind.io> * Added non-linear test with dynamic_partition. * Improved test for dynamic_partition. * dynamic_partition TAD concept Signed-off-by: raver119 <raver119@gmail.com> * - dynamic_partition TAD CUDA impl - dynamic_partition TAD CPU fix Signed-off-by: raver119 <raver119@gmail.com> * - rewrite cpu code for usampling2d/3d - write cuda code for usampling2d/3d Signed-off-by: Yurii <yurii@skymind.io> * dynamic_stitch CUDA vector case Signed-off-by: raver119 <raver119@gmail.com> * dynamic_stitch CUDA TAD case concept Signed-off-by: raver119 <raver119@gmail.com> * dynamic_stitch CUDA TAD case impl Signed-off-by: raver119 <raver119@gmail.com> * Added tests for dynamic_stitch 3D-4D cases. * minor tests tweaks Signed-off-by: raver119 <raver119@gmail.com> * Fixed type check for dynamic stitch. * min/max bp Signed-off-by: raver119 <raver119@gmail.com> * rewrite code for upsampling2d/3d cpu Signed-off-by: Yurii <yurii@skymind.io> * reduce min/max/norm_max bp Signed-off-by: raver119 <raver119@gmail.com> * lup implementation. Additional enhancements. * provide code for upsamling2d/3d backprop Signed-off-by: Yurii <yurii@skymind.io> * weightedCrossEntropyWithLogits Signed-off-by: raver119 <raver119@gmail.com> * Fixed template math atomicMul for 64bit ints. * Refactored dynamic_partition_bp op. * inverseBroadcast fix Signed-off-by: raver119 <raver119@gmail.com> * DynamicPartitionBP test datatype fixed. * - nd4j_atomicMul Windows fix - cpu/NDArrayLambda.hpp excluded from CUDA Signed-off-by: raver119 <raver119@gmail.com>
2019-06-27 17:37:04 +02:00
////////////////////////////////////////////////////////////////////////
template <>
NDArray NDArrayFactory::create<bool>(const char order, const std::vector<Nd4jLong> &shape, const std::vector<bool> &data, nd4j::LaunchContext * context) {
if ((int) shape.size() > MAX_RANK)
throw std::invalid_argument("NDArrayFactory::create: rank of NDArray can't exceed 32 !");
ShapeDescriptor descriptor(nd4j::DataType::BOOL, order, shape);
if (descriptor.arrLength() != data.size()) {
nd4j_printf("NDArrayFactory::create: data size [%i] doesn't match shape length [%lld]\n", data.size(), descriptor.arrLength());
throw std::runtime_error("NDArrayFactory::create: data size doesn't match shape");
}
bool* hostBuffer = nullptr;
ALLOCATE(hostBuffer, context->getWorkspace(), data.size(), bool);
std::copy(data.begin(), data.end(), hostBuffer);
std::shared_ptr<DataBuffer> buffer = std::make_shared<DataBuffer>(hostBuffer, data.size() * sizeof(bool), nd4j::DataType::BOOL, true, context->getWorkspace());
NDArray result(buffer, descriptor, context);
return result;
}
////////////////////////////////////////////////////////////////////////
template <typename T>
NDArray NDArrayFactory::create(const char order, const std::vector<Nd4jLong> &shape, const std::vector<T> &data, nd4j::LaunchContext * context) {
if ((int) shape.size() > MAX_RANK)
throw std::invalid_argument("NDArrayFactory::create: rank of NDArray can't exceed 32 !");
ShapeDescriptor descriptor(DataTypeUtils::fromT<T>(), order, shape);
if (descriptor.arrLength() != data.size()) {
nd4j_printf("NDArrayFactory::create: data size [%i] doesn't match shape length [%lld]\n", data.size(), descriptor.arrLength());
throw std::runtime_error("NDArrayFactory::create: data size doesn't match shape");
}
std::shared_ptr<DataBuffer> buffer = std::make_shared<DataBuffer>(data.data(), DataTypeUtils::fromT<T>(), descriptor.arrLength() * sizeof(T), context->getWorkspace());
NDArray result(buffer, descriptor, context);
return result;
}
2019-06-06 14:21:15 +02:00
NDArray NDArrayFactory::string(const char *str, nd4j::LaunchContext * context) {
std::string s(str);
return string(s, context);
}
NDArray* NDArrayFactory::string_(const char *str, nd4j::LaunchContext * context) {
return string_(std::string(str), context);
}
NDArray NDArrayFactory::string(const std::string &str, nd4j::LaunchContext * context) {
auto headerLength = ShapeUtils::stringBufferHeaderRequirements(1);
std::shared_ptr<DataBuffer> pBuffer = std::make_shared<DataBuffer>(headerLength + str.length(), DataType::UTF8, context->getWorkspace(), true);
NDArray res(pBuffer, ShapeDescriptor::scalarDescriptor(DataType::UTF8), context);
int8_t* buffer = reinterpret_cast<int8_t*>(res.getBuffer());
auto offsets = reinterpret_cast<Nd4jLong *>(buffer);
offsets[0] = 0;
offsets[1] = str.length();
auto data = buffer + headerLength;
memcpy(data, str.c_str(), str.length());
res.tickWriteHost();
res.syncToDevice();
return res;
}
NDArray* NDArrayFactory::string_(const std::string &str, nd4j::LaunchContext * context) {
auto res = new NDArray();
*res = NDArrayFactory::string(str, context);
return res;
}
////////////////////////////////////////////////////////////////////////
template<typename T>
NDArray* NDArrayFactory::create_(const char order, const std::vector<Nd4jLong> &shape, nd4j::LaunchContext * context) {
return create_(order, shape, DataTypeUtils::fromT<T>(), context);
}
BUILD_SINGLE_TEMPLATE(template NDArray* NDArrayFactory::create_, (const char order, const std::vector<Nd4jLong> &shape, nd4j::LaunchContext * context), LIBND4J_TYPES);
////////////////////////////////////////////////////////////////////////
template <typename T>
void NDArrayFactory::memcpyFromVector(void *ptr, const std::vector<T> &vector) {
memcpy(ptr, vector.data(), vector.size() * sizeof(T));
}
template <>
void NDArrayFactory::memcpyFromVector(void *ptr, const std::vector<bool> &vector) {
auto p = reinterpret_cast<bool *>(ptr);
for (Nd4jLong e = 0; e < vector.size(); e++)
p[e] = vector[e];
}
template void NDArrayFactory::memcpyFromVector(void *ptr, const std::vector<double> &vector);
template void NDArrayFactory::memcpyFromVector(void *ptr, const std::vector<float> &vector);
template void NDArrayFactory::memcpyFromVector(void *ptr, const std::vector<float16> &vector);
template void NDArrayFactory::memcpyFromVector(void *ptr, const std::vector<Nd4jLong> &vector);
template void NDArrayFactory::memcpyFromVector(void *ptr, const std::vector<int> &vector);
template void NDArrayFactory::memcpyFromVector(void *ptr, const std::vector<int16_t> &vector);
template void NDArrayFactory::memcpyFromVector(void *ptr, const std::vector<uint8_t> &vector);
template void NDArrayFactory::memcpyFromVector(void *ptr, const std::vector<int8_t> &vector);
#ifndef __JAVACPP_HACK__
////////////////////////////////////////////////////////////////////////
template <typename T>
NDArray* NDArrayFactory::valueOf(const std::initializer_list<Nd4jLong>& shape, const T value, const char order, nd4j::LaunchContext * context) {
return valueOf(std::vector<Nd4jLong>(shape), value, order);
}
template NDArray* NDArrayFactory::valueOf(const std::initializer_list<Nd4jLong>& shape, const double value, const char order, nd4j::LaunchContext * context);
template NDArray* NDArrayFactory::valueOf(const std::initializer_list<Nd4jLong>& shape, const float value, const char order, nd4j::LaunchContext * context);
template NDArray* NDArrayFactory::valueOf(const std::initializer_list<Nd4jLong>& shape, const float16 value, const char order, nd4j::LaunchContext * context);
template NDArray* NDArrayFactory::valueOf(const std::initializer_list<Nd4jLong>& shape, const bfloat16 value, const char order, nd4j::LaunchContext * context);
template NDArray* NDArrayFactory::valueOf(const std::initializer_list<Nd4jLong>& shape, const Nd4jLong value, const char order, nd4j::LaunchContext * context);
template NDArray* NDArrayFactory::valueOf(const std::initializer_list<Nd4jLong>& shape, const int value, const char order, nd4j::LaunchContext * context);
template NDArray* NDArrayFactory::valueOf(const std::initializer_list<Nd4jLong>& shape, const uint8_t value, const char order, nd4j::LaunchContext * context);
template NDArray* NDArrayFactory::valueOf(const std::initializer_list<Nd4jLong>& shape, const int8_t value, const char order, nd4j::LaunchContext * context);
template NDArray* NDArrayFactory::valueOf(const std::initializer_list<Nd4jLong>& shape, const int16_t value, const char order, nd4j::LaunchContext * context);
template NDArray* NDArrayFactory::valueOf(const std::initializer_list<Nd4jLong>& shape, const bool value, const char order, nd4j::LaunchContext * context);
////////////////////////////////////////////////////////////////////////
template <typename T>
NDArray NDArrayFactory::create(const char order, const std::vector<Nd4jLong> &shape, const std::initializer_list<T>& data, nd4j::LaunchContext * context) {
std::vector<T> vec(data);
return create<T>(order, shape, vec, context);
}
template NDArray NDArrayFactory::create(const char order, const std::vector<Nd4jLong> &shape, const std::initializer_list<double>& data, nd4j::LaunchContext * context);
template NDArray NDArrayFactory::create(const char order, const std::vector<Nd4jLong> &shape, const std::initializer_list<float>& data, nd4j::LaunchContext * context);
template NDArray NDArrayFactory::create(const char order, const std::vector<Nd4jLong> &shape, const std::initializer_list<float16>& data, nd4j::LaunchContext * context);
template NDArray NDArrayFactory::create(const char order, const std::vector<Nd4jLong> &shape, const std::initializer_list<bfloat16>& data, nd4j::LaunchContext * context);
template NDArray NDArrayFactory::create(const char order, const std::vector<Nd4jLong> &shape, const std::initializer_list<Nd4jLong>& data, nd4j::LaunchContext * context);
template NDArray NDArrayFactory::create(const char order, const std::vector<Nd4jLong> &shape, const std::initializer_list<int>& data, nd4j::LaunchContext * context);
template NDArray NDArrayFactory::create(const char order, const std::vector<Nd4jLong> &shape, const std::initializer_list<int16_t>& data, nd4j::LaunchContext * context);
template NDArray NDArrayFactory::create(const char order, const std::vector<Nd4jLong> &shape, const std::initializer_list<int8_t>& data, nd4j::LaunchContext * context);
template NDArray NDArrayFactory::create(const char order, const std::vector<Nd4jLong> &shape, const std::initializer_list<uint8_t>& data, nd4j::LaunchContext * context);
template NDArray NDArrayFactory::create(const char order, const std::vector<Nd4jLong> &shape, const std::initializer_list<bool>& data, nd4j::LaunchContext * context);
#endif
////////////////////////////////////////////////////////////////////////
template <typename T>
NDArray* NDArrayFactory::create_(const T scalar, nd4j::LaunchContext * context) {
std::shared_ptr<DataBuffer> buffer = std::make_shared<DataBuffer>(1 * sizeof(T), DataTypeUtils::fromT<T>(), context->getWorkspace(), true);
NDArray* res = new NDArray(buffer, ShapeDescriptor::scalarDescriptor(DataTypeUtils::fromT<T>()), context);
res->bufferAsT<T>()[0] = scalar;
res->tickWriteHost();
res->syncToDevice();
return res;
}
template NDArray* NDArrayFactory::create_(const double scalar, nd4j::LaunchContext * context);
template NDArray* NDArrayFactory::create_(const float scalar, nd4j::LaunchContext * context);
template NDArray* NDArrayFactory::create_(const float16 scalar, nd4j::LaunchContext * context);
template NDArray* NDArrayFactory::create_(const bfloat16 scalar, nd4j::LaunchContext * context);
template NDArray* NDArrayFactory::create_(const Nd4jLong scalar, nd4j::LaunchContext * context);
template NDArray* NDArrayFactory::create_(const int scalar, nd4j::LaunchContext * context);
template NDArray* NDArrayFactory::create_(const bool scalar, nd4j::LaunchContext * context);
template NDArray* NDArrayFactory::create_(const int8_t scalar, nd4j::LaunchContext * context);
template NDArray* NDArrayFactory::create_(const uint8_t scalar, nd4j::LaunchContext * context);
template NDArray* NDArrayFactory::create_(const int16_t scalar, nd4j::LaunchContext * context);
template <typename T>
NDArray NDArrayFactory::create(nd4j::DataType type, const T scalar, nd4j::LaunchContext * context) {
if (type == DataTypeUtils::fromT<T>())
return NDArrayFactory::create(scalar, context);
NDArray res(type, context);
res.p(0, scalar);
res.syncToDevice();
return res;
}
// BUILD_DOUBLE_TEMPLATE(template NDArray NDArrayFactory::create, (DataType type, const T scalar, nd4j::LaunchContext * context), LIBND4J_TYPES);
template NDArray NDArrayFactory::create(DataType type, const double scalar, nd4j::LaunchContext * context);
template NDArray NDArrayFactory::create(DataType type, const float scalar, nd4j::LaunchContext * context);
template NDArray NDArrayFactory::create(DataType type, const float16 scalar, nd4j::LaunchContext * context);
template NDArray NDArrayFactory::create(DataType type, const bfloat16 scalar, nd4j::LaunchContext * context);
template NDArray NDArrayFactory::create(DataType type, const Nd4jLong scalar, nd4j::LaunchContext * context);
template NDArray NDArrayFactory::create(DataType type, const int scalar, nd4j::LaunchContext * context);
template NDArray NDArrayFactory::create(DataType type, const int8_t scalar, nd4j::LaunchContext * context);
template NDArray NDArrayFactory::create(DataType type, const uint8_t scalar, nd4j::LaunchContext * context);
template NDArray NDArrayFactory::create(DataType type, const uint16_t scalar, nd4j::LaunchContext* workspace);
template NDArray NDArrayFactory::create(DataType type, const uint32_t scalar, nd4j::LaunchContext* workspace);
template NDArray NDArrayFactory::create(DataType type, const uint64_t scalar, nd4j::LaunchContext* workspace);
template NDArray NDArrayFactory::create(DataType type, const int16_t scalar, nd4j::LaunchContext * context);
template NDArray NDArrayFactory::create(DataType type, const bool scalar, nd4j::LaunchContext * context);
template <typename T>
NDArray NDArrayFactory::create(const T scalar, nd4j::LaunchContext * context) {
std::shared_ptr<DataBuffer> buffer = std::make_shared<DataBuffer>(1 * sizeof(T), DataTypeUtils::fromT<T>(), context->getWorkspace(), true);
NDArray res(buffer, ShapeDescriptor::scalarDescriptor(DataTypeUtils::fromT<T>()), context);
res.bufferAsT<T>()[0] = scalar;
res.tickWriteHost();
res.syncToDevice();
return res;
}
template NDArray NDArrayFactory::create(const double scalar, nd4j::LaunchContext * context);
template NDArray NDArrayFactory::create(const float scalar, nd4j::LaunchContext * context);
template NDArray NDArrayFactory::create(const float16 scalar, nd4j::LaunchContext * context);
template NDArray NDArrayFactory::create(const bfloat16 scalar, nd4j::LaunchContext * context);
template NDArray NDArrayFactory::create(const Nd4jLong scalar, nd4j::LaunchContext * context);
template NDArray NDArrayFactory::create(const int scalar, nd4j::LaunchContext * context);
template NDArray NDArrayFactory::create(const int8_t scalar, nd4j::LaunchContext * context);
template NDArray NDArrayFactory::create(const uint8_t scalar, nd4j::LaunchContext * context);
template NDArray NDArrayFactory::create(const int16_t scalar, nd4j::LaunchContext * context);
template NDArray NDArrayFactory::create(const uint16_t scalar, nd4j::LaunchContext* workspace);
template NDArray NDArrayFactory::create(const uint32_t scalar, nd4j::LaunchContext* workspace);
template NDArray NDArrayFactory::create(const uint64_t scalar, nd4j::LaunchContext* workspace);
template NDArray NDArrayFactory::create(const bool scalar, nd4j::LaunchContext * context);
////////////////////////////////////////////////////////////////////////
template<typename T>
NDArray* NDArrayFactory::create_(const char order, const std::vector<Nd4jLong> &shape, const std::vector<T> &data, nd4j::LaunchContext * context) {
return new NDArray(NDArrayFactory::create<T>(order, shape, data, context));
}
template NDArray* NDArrayFactory::create_(const char order, const std::vector<Nd4jLong> &shape, const std::vector<double> &data, nd4j::LaunchContext * context);
template NDArray* NDArrayFactory::create_(const char order, const std::vector<Nd4jLong> &shape, const std::vector<float> &data, nd4j::LaunchContext * context);
template NDArray* NDArrayFactory::create_(const char order, const std::vector<Nd4jLong> &shape, const std::vector<float16> &data, nd4j::LaunchContext * context);
template NDArray* NDArrayFactory::create_(const char order, const std::vector<Nd4jLong> &shape, const std::vector<bfloat16> &data, nd4j::LaunchContext * context);
template NDArray* NDArrayFactory::create_(const char order, const std::vector<Nd4jLong> &shape, const std::vector<int> &data, nd4j::LaunchContext * context);
Merge master to upstream (#7945) * Shugeo strided slice zeros (#14) * Modified strided_slice op to properly work with empty-like shapes. * Fixed test for reduce_mean with empty-like input. * [WIP] Last merge (#15) * correct logsoftmax looss (#2) * Small SameDiff listener fix (#4) * Various fixes (#6) * #7839 Fix for asXMatrix and tests * #7866 EmbeddingSequenceLayer dtype fix + test * #7856 SameDiff save/load stream methods * #7859 RegressionEvaluation rank 4 fix + tests + axis configuration * EvaluationBinary 3d/4d * More evaluation 3d/4d tests * #7847 Evaluation empty checks * Small test ifx * #7848 Fix median edge case * Improve DL4J samediff layer tests * [WIP] FastText wrapper implemented (#8) * FastText implemented * Some fixes * Fix shapes for wordsNearest * Validation of input vectors * Fixes * Fixed test * Thread tagged * Some tweaks * setContextClassLoader for DeallocatorServiceThread * Numpy format tests (#1) * Various fixes (#11) * #7852 SameDiff gather fix * #7892 SameDiff placeholder to constant conversion * #7890 validate input rank for MLN/CG init methods * Fix broken permute shape calculation * Permute and gather fixes * Tests * #7850 LogSumExp fix + test * Handful of test fixes * Empty arrays with non-scalar shapes (#10) * minor rearrangements for lambdas * empty tensors with non-scalar shapes * numpy empty tensors with non-scalar shapes * few more empty tweaks * Small fixes * conv3d signature update * micro fix in batchnorm mkldnn * Import fixes * Fix * MKL-DNN update * Small fill fix * fill with empty input + test * Fixes * Small error improvement * Fix * one special test * couple of fixes for lstm * Rewrite TFGraphMapper.getNDArrayFromTensor to be maintainable and less error prone * Fixes * FP16 * Unsigned * BFloat16 * Fill op - empty tweaks * - couple of fixes for empty arrays construction - stack updated * strided slice fix * one transform test * provide method for reducing shapeInfo in case of input array is empty * Fixed reduceAlongDimensions to use empty input properly. * couple of broadcast tests * couple of tests broadcast tests + tweak to make them pass * add check of non-empty to methods producing sub-arrays * Fixed reshapeC with zeros in shape. * complete empty check in reduce_... legacy ops * Concat and cumsum/prod * Tweak to empty shape inference on import * add empty check to the rest of reduce legacy ops * one more test * correct typo in evalReduceShapeInfoEmpty * Added tests for reduce_* ops to tests with zero shapes. * few more tests for empty reductions * Fixed strided_slice op with empty case and tests. * one more empty reduction test * Fixed strided_slice test. * add empty check to NDArray::reshapei * infOrMax * empty min/max with infinity tests * made unstack working correctly with empty arrays * few IndexReduce tests + tweaks for empty shapes * add test for empty concat * few tests fixed * Validation fix for reductions on empty shapes * Reverse fix * Reduction shape calc fixes * SameDiff.generateOutputVariable: don't use shape function to determine number of outputs * Range fix * - NDArray constructor updated for scalars/empty arrays - few tests fixed * More fixes * Empty creator fixes * concat fix * concat fix * TF import tests: allow 'both all NaN' and 'both all inf' to pass * Slice, zero fraction, and reshape fixes * transpose, gather * Zero fraction * scalar cast fix * Empty reduction axis support * few more tests fixed * Fixed input checks conforming with TF for concat op and tests. * few tests fixed * matmul scalar shape fix * Fixed checkout for data type and scalarity with concat to allow non-empty scalars with vector concats. * broadcast bool fix * few more tests * few more tests * correct evalReduceShapeInfoEmpty * argmax/argmin + tests * one more empty edge case + one more test * argmax/argmin/realdiv_bp tweaks * empty reshape test + fix * Helper fixes * Small fixes * Gather test fix * Gather test fix * Small fixes * reduce scalar zero values * scalar mean workaround * Remove debug code * along dim mean workaround * one more test * - equalsTo() tweak for empty arrays - one more test * broadcast tweaks * [WIP] Fixing outstanding issues for NLP (#9) * Avoid using not-inited objects * Test fixed. * Redundant method avoided for models like FastText * KMeans++ implementation * KMeans++ implementation * Disable parallel execution * KMeans++ * Tests * Dev branch merge (#16) * SameDiff: convertDataType and gradient check util improvements (#12) * GradCheck util improvements * StopGradient constructor + test * SameDiff: Add datatype conversion * Javadoc and add DataType.isNumerical() * Small fix * Fix SameDiff TF import test cases intermediate naming (workaround for bad default) * TFGraphTestAllHelper: check intermediates in execution order * Add missing debug listener * [WIP] lstmBlock fix + other changes (#13) - fixes lstmBlock issue - changes NDArray method reshape(), permute(), transpose() by making them return instance instead of pointer - CheckNumerics op - fixes for ReduceBool IsInfOrNan & IsFinite * Small test fix * CheckNumerics op wrapper * Fix some issues on master (#17) * Fix DataVec test issue * Fix issue with dl4j SameDiff output layer * Dtype fix for lambda layers * #7912 BertIterator dtype fix (use float32 not global default) * [WIP] Next set of CUDA stuff (#7) New CUDA implementations and improvements * bad file * Dev branch master merge (#23) * SameDiff: convertDataType and gradient check util improvements (#12) * GradCheck util improvements * StopGradient constructor + test * SameDiff: Add datatype conversion * Javadoc and add DataType.isNumerical() * Small fix * Fix SameDiff TF import test cases intermediate naming (workaround for bad default) * TFGraphTestAllHelper: check intermediates in execution order * Add missing debug listener * [WIP] lstmBlock fix + other changes (#13) - fixes lstmBlock issue - changes NDArray method reshape(), permute(), transpose() by making them return instance instead of pointer - CheckNumerics op - fixes for ReduceBool IsInfOrNan & IsFinite * Small test fix * CheckNumerics op wrapper * Compatibility of deserialization (#18) Signed-off-by: Alexander Stoyakin <alexander.stoyakin@gmail.com> * SameDiff: add activation gradient checking support for debugging (#19) * SameDiff gradient checker: first pass on activation gradient checks * Fixes + tests for activation gradient checking * Javadoc * [WIP] Some nd4j data type corrections (#20) * Adjust data type * Set correct Data type. * Size of proper data type. * fix averaged cpu load (#22) * SameDiff ops, TF import and fixes (#24) * CheckNumerics tests + fixes + misc fixes Signed-off-by: AlexDBlack <blacka101@gmail.com> * Fake quant Signed-off-by: AlexDBlack <blacka101@gmail.com> * Fixes Signed-off-by: AlexDBlack <blacka101@gmail.com> * FakeQuantWithMinMaxArgs Signed-off-by: AlexDBlack <blacka101@gmail.com> * CheckNumerics fix Signed-off-by: AlexDBlack <blacka101@gmail.com> * Fix libnd4j ALL_INTS and ALL_FLOATS declaration (uint and bfloat types) Signed-off-by: AlexDBlack <blacka101@gmail.com> * Small fix Signed-off-by: AlexDBlack <blacka101@gmail.com> * Javadoc Signed-off-by: AlexDBlack <blacka101@gmail.com> * Exception tweak Signed-off-by: AlexDBlack <blacka101@gmail.com> * fix Signed-off-by: AlexDBlack <blacka101@gmail.com> * Fix for out of scope stack allocated var use Signed-off-by: AlexDBlack <blacka101@gmail.com> * Ignores Signed-off-by: AlexDBlack <blacka101@gmail.com> * Ignore for known failing test (already logged issue) Signed-off-by: AlexDBlack <blacka101@gmail.com> * Merge upstream to fork (#25) * Add thousand-separator commas to TotalParams (#7915) * Add thousand-separator commas to TotalParams The number of parameters can be quite large, and it would help the reading of the summary printout to have the TotalParams column & values at the bottom have thousand-separator-commas in them. * Add thousand-separator commas to MultiLayerNetwork Corresponding change to MultiLayerNetwork Signed-off-by: Jxtps Jxtps <jxtps435@gmail.com> * Update contributing and issue/PR templates (#7934) Signed-off-by: AlexDBlack <blacka101@gmail.com> * Fix link to AdaDelta paper (#7942) Fix link to AdaDelta paper hosted on matthewzeiler.com Signed-off-by: Jxtps * Fixes, and ignores for known/logged failing issues (#7943) Signed-off-by: AlexDBlack <blacka101@gmail.com> * SameDiff + DL4J/SameDiff: Multiple fixes (#28) * #7919 HDF5 attribute buffer length fix Signed-off-by: AlexDBlack <blacka101@gmail.com> * #7909 Arbiter constructor exception ux improvements Signed-off-by: AlexDBlack <blacka101@gmail.com> * #7925 RNN output layer length checks Signed-off-by: AlexDBlack <blacka101@gmail.com> * #7939 Add listener for validating inputs are not incorrectly modified Signed-off-by: AlexDBlack <blacka101@gmail.com> * #7939 Integrate NonInplaceValidationListener into tests * #7844 DL4J SameDiff fixes for variable minibatch size * DL4J SameDiff fixes - ensure gradient for input placeholder is available Signed-off-by: AlexDBlack <blacka101@gmail.com> * Tweaks to ExternalErrorsFunction - use placeholders, make more robust * Another fix * More fixes * More SameDiff/DL4J fixes * Scope out scalar array creation in BaseScalarOp * Remove debug code Signed-off-by: AlexDBlack <blacka101@gmail.com> * [WIP] Final dev branch merge (#29) * SameDiff: convertDataType and gradient check util improvements (#12) * GradCheck util improvements * StopGradient constructor + test * SameDiff: Add datatype conversion * Javadoc and add DataType.isNumerical() * Small fix * Fix SameDiff TF import test cases intermediate naming (workaround for bad default) * TFGraphTestAllHelper: check intermediates in execution order * Add missing debug listener * [WIP] lstmBlock fix + other changes (#13) - fixes lstmBlock issue - changes NDArray method reshape(), permute(), transpose() by making them return instance instead of pointer - CheckNumerics op - fixes for ReduceBool IsInfOrNan & IsFinite * Small test fix * CheckNumerics op wrapper * Compatibility of deserialization (#18) Signed-off-by: Alexander Stoyakin <alexander.stoyakin@gmail.com> * SameDiff: add activation gradient checking support for debugging (#19) * SameDiff gradient checker: first pass on activation gradient checks * Fixes + tests for activation gradient checking * Javadoc * [WIP] Some nd4j data type corrections (#20) * Adjust data type * Set correct Data type. * Size of proper data type. * fix averaged cpu load (#22) * [WIP] Multiple dataset iterators (#27) * Splitting dataset into arbitrary number * Fixes * Multiple split of iterator * Test * Test * Some fixes * signature change * one more tweak Signed-off-by: raver119 <raver119@gmail.com> * one more test for sequential use of DataSetIteratorSplitter Signed-off-by: raver119 <raver119@gmail.com> * Fixes * Fixes * one more test for Alexander Signed-off-by: raver119 <raver119@gmail.com> * Some fixes * Some fixes * one more test for Alexander Signed-off-by: raver119 <raver119@gmail.com> * minor test fix Signed-off-by: raver119 <raver119@gmail.com> * Some fixes * Some fixes * couple of assertions tweaked Signed-off-by: raver119 <raver119@gmail.com> * MDS splitter test :/ Signed-off-by: raver119 <raver119@gmail.com> * Minor refactoring * Multi dataset * Some fixes * More tests * Small number of test fixes/improvements (failures on CI) (#31) Signed-off-by: AlexDBlack <blacka101@gmail.com> * [WIP] More CUDA stuff (#26) * initial commit Signed-off-by: raver119 <raver119@gmail.com> * LRN BP CUDA Signed-off-by: raver119 <raver119@gmail.com> * less memory Signed-off-by: raver119 <raver119@gmail.com> * Fixed bug with crop_and_resize op helper. * get rid of unnecessary index-calculation dunction Signed-off-by: Yurii <yurii@skymind.io> * Fixed sort with nth_element cuda-based helper. * Refactored nth_element. * Refactored nth_element op and tests. * Modified usage of dim array with sortTad routine. * Refactored main routine of helper for non_max_image_suppression op. * non_max_image_suppression op helper with cuda kernel implementation. Initial revision. * fix vol2col cuda kernel * meh Signed-off-by: raver119 <raver119@gmail.com> * topK concept Signed-off-by: raver119 <raver119@gmail.com> * unsorted topK with scanWitdh of 1 Signed-off-by: raver119 <raver119@gmail.com> * correct vol2col tests * sorted/unsorted topK Signed-off-by: raver119 <raver119@gmail.com> * implementation and fixing col2im/col2vol * Corrected usage flags with input/output with reverse op. * dup is const now Signed-off-by: raver119 <raver119@gmail.com> * percentile op Signed-off-by: raver119 <raver119@gmail.com> * group tests for mapool2d Signed-off-by: Yurii <yurii@skymind.io> * special test for george Signed-off-by: raver119 <raver119@gmail.com> * less threads for sortTad Signed-off-by: raver119 <raver119@gmail.com> * provide conv2d for cuda Signed-off-by: Yurii <yurii@skymind.io> * remove auther in sort tad kernel code Signed-off-by: Yurii <yurii@skymind.io> * provide depthwise_conv2d for cuda Signed-off-by: Yurii <yurii@skymind.io> * - max_pooling_with_argmax - null check for special use Signed-off-by: raver119 <raver119@gmail.com> * dts cuda Signed-off-by: raver119 <raver119@gmail.com> * provide sconv2d for cuda Signed-off-by: Yurii <yurii@skymind.io> * std cuda Signed-off-by: raver119 <raver119@gmail.com> * Refactored non_max_suppression op to conform TF implementation. * Improved suppression helper. * provide pooling3d for cuda Signed-off-by: Yurii <yurii@skymind.io> * minor lstm rearrangements Signed-off-by: raver119 <raver119@gmail.com> * more of minor lstm rearrangements Signed-off-by: raver119 <raver119@gmail.com> * (bi)dynamic_rnn Signed-off-by: raver119 <raver119@gmail.com> * templates init order Signed-off-by: raver119 <raver119@gmail.com> * Refactored non_max_suppression op. * Added cuda kernel for non_max_suppression. * CPU sort by key/value Signed-off-by: raver119 <raver119@gmail.com> * CPU sort TAD by key/value Signed-off-by: raver119 <raver119@gmail.com> * CPU sort TAD by key/value tests Signed-off-by: raver119 <raver119@gmail.com> * Eliminate compiler error with cuda implementation. * - repaired gradCheck in cuda - provide conv2d_bp for cuda Signed-off-by: Yurii <yurii@skymind.io> * missed signature Signed-off-by: raver119 <raver119@gmail.com> * provide depthwise_conv2d_bp for cuda Signed-off-by: Yurii <yurii@skymind.io> * Implementation of lup helper with cuda kernel. Initial commit. * further work on backprops for convolutions Signed-off-by: Yurii <yurii@skymind.io> * CUDA linear sort by key/val Signed-off-by: raver119 <raver119@gmail.com> * CUDA tad sort by key/val Signed-off-by: raver119 <raver119@gmail.com> * start providing of backprop for pooling2d/3d Signed-off-by: Yurii <yurii@skymind.io> * Added atomicAdd for bool datatype. * dynamic partition concept Signed-off-by: raver119 <raver119@gmail.com> * dynamic partition concept Signed-off-by: raver119 <raver119@gmail.com> * dynamic partition scalar CUDA Signed-off-by: raver119 <raver119@gmail.com> * important comment Signed-off-by: raver119 <raver119@gmail.com> * fix pooling2d/3d backprop helpers Signed-off-by: Yurii <yurii@skymind.io> * Added non-linear test with dynamic_partition. * Improved test for dynamic_partition. * dynamic_partition TAD concept Signed-off-by: raver119 <raver119@gmail.com> * - dynamic_partition TAD CUDA impl - dynamic_partition TAD CPU fix Signed-off-by: raver119 <raver119@gmail.com> * - rewrite cpu code for usampling2d/3d - write cuda code for usampling2d/3d Signed-off-by: Yurii <yurii@skymind.io> * dynamic_stitch CUDA vector case Signed-off-by: raver119 <raver119@gmail.com> * dynamic_stitch CUDA TAD case concept Signed-off-by: raver119 <raver119@gmail.com> * dynamic_stitch CUDA TAD case impl Signed-off-by: raver119 <raver119@gmail.com> * Added tests for dynamic_stitch 3D-4D cases. * minor tests tweaks Signed-off-by: raver119 <raver119@gmail.com> * Fixed type check for dynamic stitch. * min/max bp Signed-off-by: raver119 <raver119@gmail.com> * rewrite code for upsampling2d/3d cpu Signed-off-by: Yurii <yurii@skymind.io> * reduce min/max/norm_max bp Signed-off-by: raver119 <raver119@gmail.com> * lup implementation. Additional enhancements. * provide code for upsamling2d/3d backprop Signed-off-by: Yurii <yurii@skymind.io> * weightedCrossEntropyWithLogits Signed-off-by: raver119 <raver119@gmail.com> * Fixed template math atomicMul for 64bit ints. * Refactored dynamic_partition_bp op. * inverseBroadcast fix Signed-off-by: raver119 <raver119@gmail.com> * DynamicPartitionBP test datatype fixed. * - nd4j_atomicMul Windows fix - cpu/NDArrayLambda.hpp excluded from CUDA Signed-off-by: raver119 <raver119@gmail.com>
2019-06-27 17:37:04 +02:00
template NDArray* NDArrayFactory::create_(const char order, const std::vector<Nd4jLong> &shape, const std::vector<unsigned int> &data, nd4j::LaunchContext * context);
template NDArray* NDArrayFactory::create_(const char order, const std::vector<Nd4jLong> &shape, const std::vector<unsigned long> &data, nd4j::LaunchContext * context);
2019-06-06 14:21:15 +02:00
template NDArray* NDArrayFactory::create_(const char order, const std::vector<Nd4jLong> &shape, const std::vector<Nd4jLong> &data, nd4j::LaunchContext * context);
template NDArray* NDArrayFactory::create_(const char order, const std::vector<Nd4jLong> &shape, const std::vector<int8_t> &data, nd4j::LaunchContext * context);
template NDArray* NDArrayFactory::create_(const char order, const std::vector<Nd4jLong> &shape, const std::vector<uint8_t> &data, nd4j::LaunchContext * context);
template NDArray* NDArrayFactory::create_(const char order, const std::vector<Nd4jLong> &shape, const std::vector<int16_t> &data, nd4j::LaunchContext * context);
Merge master to upstream (#7945) * Shugeo strided slice zeros (#14) * Modified strided_slice op to properly work with empty-like shapes. * Fixed test for reduce_mean with empty-like input. * [WIP] Last merge (#15) * correct logsoftmax looss (#2) * Small SameDiff listener fix (#4) * Various fixes (#6) * #7839 Fix for asXMatrix and tests * #7866 EmbeddingSequenceLayer dtype fix + test * #7856 SameDiff save/load stream methods * #7859 RegressionEvaluation rank 4 fix + tests + axis configuration * EvaluationBinary 3d/4d * More evaluation 3d/4d tests * #7847 Evaluation empty checks * Small test ifx * #7848 Fix median edge case * Improve DL4J samediff layer tests * [WIP] FastText wrapper implemented (#8) * FastText implemented * Some fixes * Fix shapes for wordsNearest * Validation of input vectors * Fixes * Fixed test * Thread tagged * Some tweaks * setContextClassLoader for DeallocatorServiceThread * Numpy format tests (#1) * Various fixes (#11) * #7852 SameDiff gather fix * #7892 SameDiff placeholder to constant conversion * #7890 validate input rank for MLN/CG init methods * Fix broken permute shape calculation * Permute and gather fixes * Tests * #7850 LogSumExp fix + test * Handful of test fixes * Empty arrays with non-scalar shapes (#10) * minor rearrangements for lambdas * empty tensors with non-scalar shapes * numpy empty tensors with non-scalar shapes * few more empty tweaks * Small fixes * conv3d signature update * micro fix in batchnorm mkldnn * Import fixes * Fix * MKL-DNN update * Small fill fix * fill with empty input + test * Fixes * Small error improvement * Fix * one special test * couple of fixes for lstm * Rewrite TFGraphMapper.getNDArrayFromTensor to be maintainable and less error prone * Fixes * FP16 * Unsigned * BFloat16 * Fill op - empty tweaks * - couple of fixes for empty arrays construction - stack updated * strided slice fix * one transform test * provide method for reducing shapeInfo in case of input array is empty * Fixed reduceAlongDimensions to use empty input properly. * couple of broadcast tests * couple of tests broadcast tests + tweak to make them pass * add check of non-empty to methods producing sub-arrays * Fixed reshapeC with zeros in shape. * complete empty check in reduce_... legacy ops * Concat and cumsum/prod * Tweak to empty shape inference on import * add empty check to the rest of reduce legacy ops * one more test * correct typo in evalReduceShapeInfoEmpty * Added tests for reduce_* ops to tests with zero shapes. * few more tests for empty reductions * Fixed strided_slice op with empty case and tests. * one more empty reduction test * Fixed strided_slice test. * add empty check to NDArray::reshapei * infOrMax * empty min/max with infinity tests * made unstack working correctly with empty arrays * few IndexReduce tests + tweaks for empty shapes * add test for empty concat * few tests fixed * Validation fix for reductions on empty shapes * Reverse fix * Reduction shape calc fixes * SameDiff.generateOutputVariable: don't use shape function to determine number of outputs * Range fix * - NDArray constructor updated for scalars/empty arrays - few tests fixed * More fixes * Empty creator fixes * concat fix * concat fix * TF import tests: allow 'both all NaN' and 'both all inf' to pass * Slice, zero fraction, and reshape fixes * transpose, gather * Zero fraction * scalar cast fix * Empty reduction axis support * few more tests fixed * Fixed input checks conforming with TF for concat op and tests. * few tests fixed * matmul scalar shape fix * Fixed checkout for data type and scalarity with concat to allow non-empty scalars with vector concats. * broadcast bool fix * few more tests * few more tests * correct evalReduceShapeInfoEmpty * argmax/argmin + tests * one more empty edge case + one more test * argmax/argmin/realdiv_bp tweaks * empty reshape test + fix * Helper fixes * Small fixes * Gather test fix * Gather test fix * Small fixes * reduce scalar zero values * scalar mean workaround * Remove debug code * along dim mean workaround * one more test * - equalsTo() tweak for empty arrays - one more test * broadcast tweaks * [WIP] Fixing outstanding issues for NLP (#9) * Avoid using not-inited objects * Test fixed. * Redundant method avoided for models like FastText * KMeans++ implementation * KMeans++ implementation * Disable parallel execution * KMeans++ * Tests * Dev branch merge (#16) * SameDiff: convertDataType and gradient check util improvements (#12) * GradCheck util improvements * StopGradient constructor + test * SameDiff: Add datatype conversion * Javadoc and add DataType.isNumerical() * Small fix * Fix SameDiff TF import test cases intermediate naming (workaround for bad default) * TFGraphTestAllHelper: check intermediates in execution order * Add missing debug listener * [WIP] lstmBlock fix + other changes (#13) - fixes lstmBlock issue - changes NDArray method reshape(), permute(), transpose() by making them return instance instead of pointer - CheckNumerics op - fixes for ReduceBool IsInfOrNan & IsFinite * Small test fix * CheckNumerics op wrapper * Fix some issues on master (#17) * Fix DataVec test issue * Fix issue with dl4j SameDiff output layer * Dtype fix for lambda layers * #7912 BertIterator dtype fix (use float32 not global default) * [WIP] Next set of CUDA stuff (#7) New CUDA implementations and improvements * bad file * Dev branch master merge (#23) * SameDiff: convertDataType and gradient check util improvements (#12) * GradCheck util improvements * StopGradient constructor + test * SameDiff: Add datatype conversion * Javadoc and add DataType.isNumerical() * Small fix * Fix SameDiff TF import test cases intermediate naming (workaround for bad default) * TFGraphTestAllHelper: check intermediates in execution order * Add missing debug listener * [WIP] lstmBlock fix + other changes (#13) - fixes lstmBlock issue - changes NDArray method reshape(), permute(), transpose() by making them return instance instead of pointer - CheckNumerics op - fixes for ReduceBool IsInfOrNan & IsFinite * Small test fix * CheckNumerics op wrapper * Compatibility of deserialization (#18) Signed-off-by: Alexander Stoyakin <alexander.stoyakin@gmail.com> * SameDiff: add activation gradient checking support for debugging (#19) * SameDiff gradient checker: first pass on activation gradient checks * Fixes + tests for activation gradient checking * Javadoc * [WIP] Some nd4j data type corrections (#20) * Adjust data type * Set correct Data type. * Size of proper data type. * fix averaged cpu load (#22) * SameDiff ops, TF import and fixes (#24) * CheckNumerics tests + fixes + misc fixes Signed-off-by: AlexDBlack <blacka101@gmail.com> * Fake quant Signed-off-by: AlexDBlack <blacka101@gmail.com> * Fixes Signed-off-by: AlexDBlack <blacka101@gmail.com> * FakeQuantWithMinMaxArgs Signed-off-by: AlexDBlack <blacka101@gmail.com> * CheckNumerics fix Signed-off-by: AlexDBlack <blacka101@gmail.com> * Fix libnd4j ALL_INTS and ALL_FLOATS declaration (uint and bfloat types) Signed-off-by: AlexDBlack <blacka101@gmail.com> * Small fix Signed-off-by: AlexDBlack <blacka101@gmail.com> * Javadoc Signed-off-by: AlexDBlack <blacka101@gmail.com> * Exception tweak Signed-off-by: AlexDBlack <blacka101@gmail.com> * fix Signed-off-by: AlexDBlack <blacka101@gmail.com> * Fix for out of scope stack allocated var use Signed-off-by: AlexDBlack <blacka101@gmail.com> * Ignores Signed-off-by: AlexDBlack <blacka101@gmail.com> * Ignore for known failing test (already logged issue) Signed-off-by: AlexDBlack <blacka101@gmail.com> * Merge upstream to fork (#25) * Add thousand-separator commas to TotalParams (#7915) * Add thousand-separator commas to TotalParams The number of parameters can be quite large, and it would help the reading of the summary printout to have the TotalParams column & values at the bottom have thousand-separator-commas in them. * Add thousand-separator commas to MultiLayerNetwork Corresponding change to MultiLayerNetwork Signed-off-by: Jxtps Jxtps <jxtps435@gmail.com> * Update contributing and issue/PR templates (#7934) Signed-off-by: AlexDBlack <blacka101@gmail.com> * Fix link to AdaDelta paper (#7942) Fix link to AdaDelta paper hosted on matthewzeiler.com Signed-off-by: Jxtps * Fixes, and ignores for known/logged failing issues (#7943) Signed-off-by: AlexDBlack <blacka101@gmail.com> * SameDiff + DL4J/SameDiff: Multiple fixes (#28) * #7919 HDF5 attribute buffer length fix Signed-off-by: AlexDBlack <blacka101@gmail.com> * #7909 Arbiter constructor exception ux improvements Signed-off-by: AlexDBlack <blacka101@gmail.com> * #7925 RNN output layer length checks Signed-off-by: AlexDBlack <blacka101@gmail.com> * #7939 Add listener for validating inputs are not incorrectly modified Signed-off-by: AlexDBlack <blacka101@gmail.com> * #7939 Integrate NonInplaceValidationListener into tests * #7844 DL4J SameDiff fixes for variable minibatch size * DL4J SameDiff fixes - ensure gradient for input placeholder is available Signed-off-by: AlexDBlack <blacka101@gmail.com> * Tweaks to ExternalErrorsFunction - use placeholders, make more robust * Another fix * More fixes * More SameDiff/DL4J fixes * Scope out scalar array creation in BaseScalarOp * Remove debug code Signed-off-by: AlexDBlack <blacka101@gmail.com> * [WIP] Final dev branch merge (#29) * SameDiff: convertDataType and gradient check util improvements (#12) * GradCheck util improvements * StopGradient constructor + test * SameDiff: Add datatype conversion * Javadoc and add DataType.isNumerical() * Small fix * Fix SameDiff TF import test cases intermediate naming (workaround for bad default) * TFGraphTestAllHelper: check intermediates in execution order * Add missing debug listener * [WIP] lstmBlock fix + other changes (#13) - fixes lstmBlock issue - changes NDArray method reshape(), permute(), transpose() by making them return instance instead of pointer - CheckNumerics op - fixes for ReduceBool IsInfOrNan & IsFinite * Small test fix * CheckNumerics op wrapper * Compatibility of deserialization (#18) Signed-off-by: Alexander Stoyakin <alexander.stoyakin@gmail.com> * SameDiff: add activation gradient checking support for debugging (#19) * SameDiff gradient checker: first pass on activation gradient checks * Fixes + tests for activation gradient checking * Javadoc * [WIP] Some nd4j data type corrections (#20) * Adjust data type * Set correct Data type. * Size of proper data type. * fix averaged cpu load (#22) * [WIP] Multiple dataset iterators (#27) * Splitting dataset into arbitrary number * Fixes * Multiple split of iterator * Test * Test * Some fixes * signature change * one more tweak Signed-off-by: raver119 <raver119@gmail.com> * one more test for sequential use of DataSetIteratorSplitter Signed-off-by: raver119 <raver119@gmail.com> * Fixes * Fixes * one more test for Alexander Signed-off-by: raver119 <raver119@gmail.com> * Some fixes * Some fixes * one more test for Alexander Signed-off-by: raver119 <raver119@gmail.com> * minor test fix Signed-off-by: raver119 <raver119@gmail.com> * Some fixes * Some fixes * couple of assertions tweaked Signed-off-by: raver119 <raver119@gmail.com> * MDS splitter test :/ Signed-off-by: raver119 <raver119@gmail.com> * Minor refactoring * Multi dataset * Some fixes * More tests * Small number of test fixes/improvements (failures on CI) (#31) Signed-off-by: AlexDBlack <blacka101@gmail.com> * [WIP] More CUDA stuff (#26) * initial commit Signed-off-by: raver119 <raver119@gmail.com> * LRN BP CUDA Signed-off-by: raver119 <raver119@gmail.com> * less memory Signed-off-by: raver119 <raver119@gmail.com> * Fixed bug with crop_and_resize op helper. * get rid of unnecessary index-calculation dunction Signed-off-by: Yurii <yurii@skymind.io> * Fixed sort with nth_element cuda-based helper. * Refactored nth_element. * Refactored nth_element op and tests. * Modified usage of dim array with sortTad routine. * Refactored main routine of helper for non_max_image_suppression op. * non_max_image_suppression op helper with cuda kernel implementation. Initial revision. * fix vol2col cuda kernel * meh Signed-off-by: raver119 <raver119@gmail.com> * topK concept Signed-off-by: raver119 <raver119@gmail.com> * unsorted topK with scanWitdh of 1 Signed-off-by: raver119 <raver119@gmail.com> * correct vol2col tests * sorted/unsorted topK Signed-off-by: raver119 <raver119@gmail.com> * implementation and fixing col2im/col2vol * Corrected usage flags with input/output with reverse op. * dup is const now Signed-off-by: raver119 <raver119@gmail.com> * percentile op Signed-off-by: raver119 <raver119@gmail.com> * group tests for mapool2d Signed-off-by: Yurii <yurii@skymind.io> * special test for george Signed-off-by: raver119 <raver119@gmail.com> * less threads for sortTad Signed-off-by: raver119 <raver119@gmail.com> * provide conv2d for cuda Signed-off-by: Yurii <yurii@skymind.io> * remove auther in sort tad kernel code Signed-off-by: Yurii <yurii@skymind.io> * provide depthwise_conv2d for cuda Signed-off-by: Yurii <yurii@skymind.io> * - max_pooling_with_argmax - null check for special use Signed-off-by: raver119 <raver119@gmail.com> * dts cuda Signed-off-by: raver119 <raver119@gmail.com> * provide sconv2d for cuda Signed-off-by: Yurii <yurii@skymind.io> * std cuda Signed-off-by: raver119 <raver119@gmail.com> * Refactored non_max_suppression op to conform TF implementation. * Improved suppression helper. * provide pooling3d for cuda Signed-off-by: Yurii <yurii@skymind.io> * minor lstm rearrangements Signed-off-by: raver119 <raver119@gmail.com> * more of minor lstm rearrangements Signed-off-by: raver119 <raver119@gmail.com> * (bi)dynamic_rnn Signed-off-by: raver119 <raver119@gmail.com> * templates init order Signed-off-by: raver119 <raver119@gmail.com> * Refactored non_max_suppression op. * Added cuda kernel for non_max_suppression. * CPU sort by key/value Signed-off-by: raver119 <raver119@gmail.com> * CPU sort TAD by key/value Signed-off-by: raver119 <raver119@gmail.com> * CPU sort TAD by key/value tests Signed-off-by: raver119 <raver119@gmail.com> * Eliminate compiler error with cuda implementation. * - repaired gradCheck in cuda - provide conv2d_bp for cuda Signed-off-by: Yurii <yurii@skymind.io> * missed signature Signed-off-by: raver119 <raver119@gmail.com> * provide depthwise_conv2d_bp for cuda Signed-off-by: Yurii <yurii@skymind.io> * Implementation of lup helper with cuda kernel. Initial commit. * further work on backprops for convolutions Signed-off-by: Yurii <yurii@skymind.io> * CUDA linear sort by key/val Signed-off-by: raver119 <raver119@gmail.com> * CUDA tad sort by key/val Signed-off-by: raver119 <raver119@gmail.com> * start providing of backprop for pooling2d/3d Signed-off-by: Yurii <yurii@skymind.io> * Added atomicAdd for bool datatype. * dynamic partition concept Signed-off-by: raver119 <raver119@gmail.com> * dynamic partition concept Signed-off-by: raver119 <raver119@gmail.com> * dynamic partition scalar CUDA Signed-off-by: raver119 <raver119@gmail.com> * important comment Signed-off-by: raver119 <raver119@gmail.com> * fix pooling2d/3d backprop helpers Signed-off-by: Yurii <yurii@skymind.io> * Added non-linear test with dynamic_partition. * Improved test for dynamic_partition. * dynamic_partition TAD concept Signed-off-by: raver119 <raver119@gmail.com> * - dynamic_partition TAD CUDA impl - dynamic_partition TAD CPU fix Signed-off-by: raver119 <raver119@gmail.com> * - rewrite cpu code for usampling2d/3d - write cuda code for usampling2d/3d Signed-off-by: Yurii <yurii@skymind.io> * dynamic_stitch CUDA vector case Signed-off-by: raver119 <raver119@gmail.com> * dynamic_stitch CUDA TAD case concept Signed-off-by: raver119 <raver119@gmail.com> * dynamic_stitch CUDA TAD case impl Signed-off-by: raver119 <raver119@gmail.com> * Added tests for dynamic_stitch 3D-4D cases. * minor tests tweaks Signed-off-by: raver119 <raver119@gmail.com> * Fixed type check for dynamic stitch. * min/max bp Signed-off-by: raver119 <raver119@gmail.com> * rewrite code for upsampling2d/3d cpu Signed-off-by: Yurii <yurii@skymind.io> * reduce min/max/norm_max bp Signed-off-by: raver119 <raver119@gmail.com> * lup implementation. Additional enhancements. * provide code for upsamling2d/3d backprop Signed-off-by: Yurii <yurii@skymind.io> * weightedCrossEntropyWithLogits Signed-off-by: raver119 <raver119@gmail.com> * Fixed template math atomicMul for 64bit ints. * Refactored dynamic_partition_bp op. * inverseBroadcast fix Signed-off-by: raver119 <raver119@gmail.com> * DynamicPartitionBP test datatype fixed. * - nd4j_atomicMul Windows fix - cpu/NDArrayLambda.hpp excluded from CUDA Signed-off-by: raver119 <raver119@gmail.com>
2019-06-27 17:37:04 +02:00
template NDArray* NDArrayFactory::create_(const char order, const std::vector<Nd4jLong> &shape, const std::vector<uint16_t> &data, nd4j::LaunchContext * context);
2019-06-06 14:21:15 +02:00
template NDArray* NDArrayFactory::create_(const char order, const std::vector<Nd4jLong> &shape, const std::vector<bool> &data, nd4j::LaunchContext * context);
////////////////////////////////////////////////////////////////////////
template <>
NDArray* NDArrayFactory::valueOf(const std::vector<Nd4jLong>& shape, NDArray* value, const char order, nd4j::LaunchContext * context) {
auto result = create_(order, shape, value->dataType(), context);
result->assign(*value);
return result;
}
template <>
NDArray* NDArrayFactory::valueOf(const std::vector<Nd4jLong>& shape, NDArray& value, const char order, nd4j::LaunchContext * context) {
auto result = create_(order, shape, value.dataType(), context);
result->assign(value);
return result;
}
template <typename T>
NDArray* NDArrayFactory::valueOf(const std::vector<Nd4jLong>& shape, const T value, const char order, nd4j::LaunchContext * context) {
auto result = create_(order, shape, DataTypeUtils::fromT<T>());
result->assign(value);
return result;
}
template NDArray* NDArrayFactory::valueOf(const std::vector<Nd4jLong>& shape, const double value, const char order, nd4j::LaunchContext * context);
template NDArray* NDArrayFactory::valueOf(const std::vector<Nd4jLong>& shape, const float value, const char order, nd4j::LaunchContext * context);
template NDArray* NDArrayFactory::valueOf(const std::vector<Nd4jLong>& shape, const float16 value, const char order, nd4j::LaunchContext * context);
template NDArray* NDArrayFactory::valueOf(const std::vector<Nd4jLong>& shape, const bfloat16 value, const char order, nd4j::LaunchContext * context);
template NDArray* NDArrayFactory::valueOf(const std::vector<Nd4jLong>& shape, const Nd4jLong value, const char order, nd4j::LaunchContext * context);
template NDArray* NDArrayFactory::valueOf(const std::vector<Nd4jLong>& shape, const int value, const char order, nd4j::LaunchContext * context);
template NDArray* NDArrayFactory::valueOf(const std::vector<Nd4jLong>& shape, const int16_t value, const char order, nd4j::LaunchContext * context);
template NDArray* NDArrayFactory::valueOf(const std::vector<Nd4jLong>& shape, const int8_t value, const char order, nd4j::LaunchContext * context);
template NDArray* NDArrayFactory::valueOf(const std::vector<Nd4jLong>& shape, const uint8_t value, const char order, nd4j::LaunchContext * context);
template NDArray* NDArrayFactory::valueOf(const std::vector<Nd4jLong>& shape, const bool value, const char order, nd4j::LaunchContext * context);
////////////////////////////////////////////////////////////////////////
template <typename T>
NDArray* NDArrayFactory::linspace(const T from, const T to, const Nd4jLong numElements) {
NDArray* result = NDArrayFactory::vector<T>(numElements);
//TO DO: linspace should be executed on DEVICE, but only CPU version implemnted!
for (Nd4jLong e = 0; e < numElements; e++) {
T step = (T) e / ((T) numElements - (T) 1);
result->p<T >(e, (from * ((T) 1 - step) + step * to));
}
result->syncToDevice();
return result;
}
template NDArray* NDArrayFactory::linspace(const double from, const double to, const Nd4jLong numElements);
template NDArray* NDArrayFactory::linspace(const float from, const float to, const Nd4jLong numElements);
template NDArray* NDArrayFactory::linspace(const float16 from, const float16 to, const Nd4jLong numElements);
template NDArray* NDArrayFactory::linspace(const bfloat16 from, const bfloat16 to, const Nd4jLong numElements);
template NDArray* NDArrayFactory::linspace(const Nd4jLong from, const Nd4jLong to, const Nd4jLong numElements);
template NDArray* NDArrayFactory::linspace(const int from, const int to, const Nd4jLong numElements);
template NDArray* NDArrayFactory::linspace(const int16_t from, const int16_t to, const Nd4jLong numElements);
template NDArray* NDArrayFactory::linspace(const uint8_t from, const uint8_t to, const Nd4jLong numElements);
template NDArray* NDArrayFactory::linspace(const uint16_t from, const uint16_t to, const Nd4jLong numElements);
template NDArray* NDArrayFactory::linspace(const uint32_t from, const uint32_t to, const Nd4jLong numElements);
template NDArray* NDArrayFactory::linspace(const uint64_t from, const uint64_t to, const Nd4jLong numElements);
template NDArray* NDArrayFactory::linspace(const int8_t from, const int8_t to, const Nd4jLong numElements);
template NDArray* NDArrayFactory::linspace(const bool from, const bool to, const Nd4jLong numElements);
////////////////////////////////////////////////////////////////////////
template <typename T>
NDArray* NDArrayFactory::vector(Nd4jLong length, const T value, nd4j::LaunchContext * context) {
std::shared_ptr<DataBuffer> buffer = std::make_shared<DataBuffer>(length * sizeof(T), DataTypeUtils::fromT<T>(), context->getWorkspace(), true);
auto res = new NDArray(buffer, ShapeDescriptor::vectorDescriptor(length, DataTypeUtils::fromT<T>()), context);
if (value == (T)0.0f)
res->nullify();
else
res->assign(value);
return res;
}
template NDArray* NDArrayFactory::vector(Nd4jLong length, const double startingValue, nd4j::LaunchContext * context);
template NDArray* NDArrayFactory::vector(Nd4jLong length, const float startingValue, nd4j::LaunchContext * context);
template NDArray* NDArrayFactory::vector(Nd4jLong length, const float16 startingValue, nd4j::LaunchContext * context);
template NDArray* NDArrayFactory::vector(Nd4jLong length, const bfloat16 startingValue, nd4j::LaunchContext * context);
template NDArray* NDArrayFactory::vector(Nd4jLong length, const Nd4jLong startingValue, nd4j::LaunchContext * context);
template NDArray* NDArrayFactory::vector(Nd4jLong length, const int startingValue, nd4j::LaunchContext * context);
template NDArray* NDArrayFactory::vector(Nd4jLong length, const uint8_t startingValue, nd4j::LaunchContext * context);
template NDArray* NDArrayFactory::vector(Nd4jLong length, const uint16_t startingValue, nd4j::LaunchContext *workspace);
template NDArray* NDArrayFactory::vector(Nd4jLong length, const uint32_t startingValue, nd4j::LaunchContext *workspace);
template NDArray* NDArrayFactory::vector(Nd4jLong length, const uint64_t startingValue, nd4j::LaunchContext *workspace);
template NDArray* NDArrayFactory::vector(Nd4jLong length, const int8_t startingValue, nd4j::LaunchContext * context);
template NDArray* NDArrayFactory::vector(Nd4jLong length, const int16_t startingValue, nd4j::LaunchContext * context);
template NDArray* NDArrayFactory::vector(Nd4jLong length, const bool startingValue, nd4j::LaunchContext * context);
////////////////////////////////////////////////////////////////////////
template <typename T>
NDArray NDArrayFactory::create(const char order, const std::initializer_list<Nd4jLong>& shape, nd4j::LaunchContext * context) {
std::vector<Nd4jLong> vec(shape);
return create<T>(order, vec, context);
}
BUILD_SINGLE_TEMPLATE(template NDArray NDArrayFactory::create, (const char, const std::initializer_list<Nd4jLong>&, nd4j::LaunchContext * context), LIBND4J_TYPES);
////////////////////////////////////////////////////////////////////////
template <typename T>
NDArray NDArrayFactory::create(const char order, const std::vector<Nd4jLong> &shape, nd4j::LaunchContext * context) {
return create(order, shape, DataTypeUtils::fromT<T>(), context);
}
BUILD_SINGLE_TEMPLATE(template NDArray NDArrayFactory::create, (const char order, const std::vector<Nd4jLong> &shape, nd4j::LaunchContext * context), LIBND4J_TYPES);
////////////////////////////////////////////////////////////////////////
NDArray NDArrayFactory::create(const char order, const std::vector<Nd4jLong> &shape, nd4j::DataType dtype, nd4j::LaunchContext* context) {
if ((int) shape.size() > MAX_RANK)
throw std::invalid_argument("NDArrayFactory::create: rank of NDArray can't exceed 32");
ShapeDescriptor descriptor(dtype, order, shape);
std::shared_ptr<DataBuffer> buffer = std::make_shared<DataBuffer>(descriptor.arrLength() * DataTypeUtils::sizeOfElement(dtype), dtype, context->getWorkspace());
NDArray result(buffer, descriptor, context);
result.nullify();
return result;
}
////////////////////////////////////////////////////////////////////////
NDArray NDArrayFactory::create(nd4j::DataType dtype, nd4j::LaunchContext * context) {
std::shared_ptr<DataBuffer> buffer = std::make_shared<DataBuffer>(DataTypeUtils::sizeOfElement(dtype), dtype, context->getWorkspace(), true);
NDArray res(buffer, ShapeDescriptor::scalarDescriptor(dtype), context);
res.nullify();
return res;
}
////////////////////////////////////////////////////////////////////////
template <typename T>
NDArray NDArrayFactory::create(const std::vector<T> &values, nd4j::LaunchContext * context) {
std::shared_ptr<DataBuffer> buffer = std::make_shared<DataBuffer>(values.size() * sizeof(T), DataTypeUtils::fromT<T>(), context->getWorkspace(), true);
NDArray res(buffer, ShapeDescriptor::vectorDescriptor(values.size(), DataTypeUtils::fromT<T>()), context);
memcpyFromVector<T>(res.getBuffer(), values);
res.tickWriteHost();
res.syncToDevice();
return res;
}
template NDArray NDArrayFactory::create(const std::vector<double> &values, nd4j::LaunchContext * context);
template NDArray NDArrayFactory::create(const std::vector<float> &values, nd4j::LaunchContext * context);
template NDArray NDArrayFactory::create(const std::vector<float16> &values, nd4j::LaunchContext * context);
template NDArray NDArrayFactory::create(const std::vector<bfloat16> &values, nd4j::LaunchContext * context);
template NDArray NDArrayFactory::create(const std::vector<Nd4jLong> &values, nd4j::LaunchContext * context);
template NDArray NDArrayFactory::create(const std::vector<int> &values, nd4j::LaunchContext * context);
template NDArray NDArrayFactory::create(const std::vector<int16_t> &values, nd4j::LaunchContext * context);
Merge master to upstream (#7945) * Shugeo strided slice zeros (#14) * Modified strided_slice op to properly work with empty-like shapes. * Fixed test for reduce_mean with empty-like input. * [WIP] Last merge (#15) * correct logsoftmax looss (#2) * Small SameDiff listener fix (#4) * Various fixes (#6) * #7839 Fix for asXMatrix and tests * #7866 EmbeddingSequenceLayer dtype fix + test * #7856 SameDiff save/load stream methods * #7859 RegressionEvaluation rank 4 fix + tests + axis configuration * EvaluationBinary 3d/4d * More evaluation 3d/4d tests * #7847 Evaluation empty checks * Small test ifx * #7848 Fix median edge case * Improve DL4J samediff layer tests * [WIP] FastText wrapper implemented (#8) * FastText implemented * Some fixes * Fix shapes for wordsNearest * Validation of input vectors * Fixes * Fixed test * Thread tagged * Some tweaks * setContextClassLoader for DeallocatorServiceThread * Numpy format tests (#1) * Various fixes (#11) * #7852 SameDiff gather fix * #7892 SameDiff placeholder to constant conversion * #7890 validate input rank for MLN/CG init methods * Fix broken permute shape calculation * Permute and gather fixes * Tests * #7850 LogSumExp fix + test * Handful of test fixes * Empty arrays with non-scalar shapes (#10) * minor rearrangements for lambdas * empty tensors with non-scalar shapes * numpy empty tensors with non-scalar shapes * few more empty tweaks * Small fixes * conv3d signature update * micro fix in batchnorm mkldnn * Import fixes * Fix * MKL-DNN update * Small fill fix * fill with empty input + test * Fixes * Small error improvement * Fix * one special test * couple of fixes for lstm * Rewrite TFGraphMapper.getNDArrayFromTensor to be maintainable and less error prone * Fixes * FP16 * Unsigned * BFloat16 * Fill op - empty tweaks * - couple of fixes for empty arrays construction - stack updated * strided slice fix * one transform test * provide method for reducing shapeInfo in case of input array is empty * Fixed reduceAlongDimensions to use empty input properly. * couple of broadcast tests * couple of tests broadcast tests + tweak to make them pass * add check of non-empty to methods producing sub-arrays * Fixed reshapeC with zeros in shape. * complete empty check in reduce_... legacy ops * Concat and cumsum/prod * Tweak to empty shape inference on import * add empty check to the rest of reduce legacy ops * one more test * correct typo in evalReduceShapeInfoEmpty * Added tests for reduce_* ops to tests with zero shapes. * few more tests for empty reductions * Fixed strided_slice op with empty case and tests. * one more empty reduction test * Fixed strided_slice test. * add empty check to NDArray::reshapei * infOrMax * empty min/max with infinity tests * made unstack working correctly with empty arrays * few IndexReduce tests + tweaks for empty shapes * add test for empty concat * few tests fixed * Validation fix for reductions on empty shapes * Reverse fix * Reduction shape calc fixes * SameDiff.generateOutputVariable: don't use shape function to determine number of outputs * Range fix * - NDArray constructor updated for scalars/empty arrays - few tests fixed * More fixes * Empty creator fixes * concat fix * concat fix * TF import tests: allow 'both all NaN' and 'both all inf' to pass * Slice, zero fraction, and reshape fixes * transpose, gather * Zero fraction * scalar cast fix * Empty reduction axis support * few more tests fixed * Fixed input checks conforming with TF for concat op and tests. * few tests fixed * matmul scalar shape fix * Fixed checkout for data type and scalarity with concat to allow non-empty scalars with vector concats. * broadcast bool fix * few more tests * few more tests * correct evalReduceShapeInfoEmpty * argmax/argmin + tests * one more empty edge case + one more test * argmax/argmin/realdiv_bp tweaks * empty reshape test + fix * Helper fixes * Small fixes * Gather test fix * Gather test fix * Small fixes * reduce scalar zero values * scalar mean workaround * Remove debug code * along dim mean workaround * one more test * - equalsTo() tweak for empty arrays - one more test * broadcast tweaks * [WIP] Fixing outstanding issues for NLP (#9) * Avoid using not-inited objects * Test fixed. * Redundant method avoided for models like FastText * KMeans++ implementation * KMeans++ implementation * Disable parallel execution * KMeans++ * Tests * Dev branch merge (#16) * SameDiff: convertDataType and gradient check util improvements (#12) * GradCheck util improvements * StopGradient constructor + test * SameDiff: Add datatype conversion * Javadoc and add DataType.isNumerical() * Small fix * Fix SameDiff TF import test cases intermediate naming (workaround for bad default) * TFGraphTestAllHelper: check intermediates in execution order * Add missing debug listener * [WIP] lstmBlock fix + other changes (#13) - fixes lstmBlock issue - changes NDArray method reshape(), permute(), transpose() by making them return instance instead of pointer - CheckNumerics op - fixes for ReduceBool IsInfOrNan & IsFinite * Small test fix * CheckNumerics op wrapper * Fix some issues on master (#17) * Fix DataVec test issue * Fix issue with dl4j SameDiff output layer * Dtype fix for lambda layers * #7912 BertIterator dtype fix (use float32 not global default) * [WIP] Next set of CUDA stuff (#7) New CUDA implementations and improvements * bad file * Dev branch master merge (#23) * SameDiff: convertDataType and gradient check util improvements (#12) * GradCheck util improvements * StopGradient constructor + test * SameDiff: Add datatype conversion * Javadoc and add DataType.isNumerical() * Small fix * Fix SameDiff TF import test cases intermediate naming (workaround for bad default) * TFGraphTestAllHelper: check intermediates in execution order * Add missing debug listener * [WIP] lstmBlock fix + other changes (#13) - fixes lstmBlock issue - changes NDArray method reshape(), permute(), transpose() by making them return instance instead of pointer - CheckNumerics op - fixes for ReduceBool IsInfOrNan & IsFinite * Small test fix * CheckNumerics op wrapper * Compatibility of deserialization (#18) Signed-off-by: Alexander Stoyakin <alexander.stoyakin@gmail.com> * SameDiff: add activation gradient checking support for debugging (#19) * SameDiff gradient checker: first pass on activation gradient checks * Fixes + tests for activation gradient checking * Javadoc * [WIP] Some nd4j data type corrections (#20) * Adjust data type * Set correct Data type. * Size of proper data type. * fix averaged cpu load (#22) * SameDiff ops, TF import and fixes (#24) * CheckNumerics tests + fixes + misc fixes Signed-off-by: AlexDBlack <blacka101@gmail.com> * Fake quant Signed-off-by: AlexDBlack <blacka101@gmail.com> * Fixes Signed-off-by: AlexDBlack <blacka101@gmail.com> * FakeQuantWithMinMaxArgs Signed-off-by: AlexDBlack <blacka101@gmail.com> * CheckNumerics fix Signed-off-by: AlexDBlack <blacka101@gmail.com> * Fix libnd4j ALL_INTS and ALL_FLOATS declaration (uint and bfloat types) Signed-off-by: AlexDBlack <blacka101@gmail.com> * Small fix Signed-off-by: AlexDBlack <blacka101@gmail.com> * Javadoc Signed-off-by: AlexDBlack <blacka101@gmail.com> * Exception tweak Signed-off-by: AlexDBlack <blacka101@gmail.com> * fix Signed-off-by: AlexDBlack <blacka101@gmail.com> * Fix for out of scope stack allocated var use Signed-off-by: AlexDBlack <blacka101@gmail.com> * Ignores Signed-off-by: AlexDBlack <blacka101@gmail.com> * Ignore for known failing test (already logged issue) Signed-off-by: AlexDBlack <blacka101@gmail.com> * Merge upstream to fork (#25) * Add thousand-separator commas to TotalParams (#7915) * Add thousand-separator commas to TotalParams The number of parameters can be quite large, and it would help the reading of the summary printout to have the TotalParams column & values at the bottom have thousand-separator-commas in them. * Add thousand-separator commas to MultiLayerNetwork Corresponding change to MultiLayerNetwork Signed-off-by: Jxtps Jxtps <jxtps435@gmail.com> * Update contributing and issue/PR templates (#7934) Signed-off-by: AlexDBlack <blacka101@gmail.com> * Fix link to AdaDelta paper (#7942) Fix link to AdaDelta paper hosted on matthewzeiler.com Signed-off-by: Jxtps * Fixes, and ignores for known/logged failing issues (#7943) Signed-off-by: AlexDBlack <blacka101@gmail.com> * SameDiff + DL4J/SameDiff: Multiple fixes (#28) * #7919 HDF5 attribute buffer length fix Signed-off-by: AlexDBlack <blacka101@gmail.com> * #7909 Arbiter constructor exception ux improvements Signed-off-by: AlexDBlack <blacka101@gmail.com> * #7925 RNN output layer length checks Signed-off-by: AlexDBlack <blacka101@gmail.com> * #7939 Add listener for validating inputs are not incorrectly modified Signed-off-by: AlexDBlack <blacka101@gmail.com> * #7939 Integrate NonInplaceValidationListener into tests * #7844 DL4J SameDiff fixes for variable minibatch size * DL4J SameDiff fixes - ensure gradient for input placeholder is available Signed-off-by: AlexDBlack <blacka101@gmail.com> * Tweaks to ExternalErrorsFunction - use placeholders, make more robust * Another fix * More fixes * More SameDiff/DL4J fixes * Scope out scalar array creation in BaseScalarOp * Remove debug code Signed-off-by: AlexDBlack <blacka101@gmail.com> * [WIP] Final dev branch merge (#29) * SameDiff: convertDataType and gradient check util improvements (#12) * GradCheck util improvements * StopGradient constructor + test * SameDiff: Add datatype conversion * Javadoc and add DataType.isNumerical() * Small fix * Fix SameDiff TF import test cases intermediate naming (workaround for bad default) * TFGraphTestAllHelper: check intermediates in execution order * Add missing debug listener * [WIP] lstmBlock fix + other changes (#13) - fixes lstmBlock issue - changes NDArray method reshape(), permute(), transpose() by making them return instance instead of pointer - CheckNumerics op - fixes for ReduceBool IsInfOrNan & IsFinite * Small test fix * CheckNumerics op wrapper * Compatibility of deserialization (#18) Signed-off-by: Alexander Stoyakin <alexander.stoyakin@gmail.com> * SameDiff: add activation gradient checking support for debugging (#19) * SameDiff gradient checker: first pass on activation gradient checks * Fixes + tests for activation gradient checking * Javadoc * [WIP] Some nd4j data type corrections (#20) * Adjust data type * Set correct Data type. * Size of proper data type. * fix averaged cpu load (#22) * [WIP] Multiple dataset iterators (#27) * Splitting dataset into arbitrary number * Fixes * Multiple split of iterator * Test * Test * Some fixes * signature change * one more tweak Signed-off-by: raver119 <raver119@gmail.com> * one more test for sequential use of DataSetIteratorSplitter Signed-off-by: raver119 <raver119@gmail.com> * Fixes * Fixes * one more test for Alexander Signed-off-by: raver119 <raver119@gmail.com> * Some fixes * Some fixes * one more test for Alexander Signed-off-by: raver119 <raver119@gmail.com> * minor test fix Signed-off-by: raver119 <raver119@gmail.com> * Some fixes * Some fixes * couple of assertions tweaked Signed-off-by: raver119 <raver119@gmail.com> * MDS splitter test :/ Signed-off-by: raver119 <raver119@gmail.com> * Minor refactoring * Multi dataset * Some fixes * More tests * Small number of test fixes/improvements (failures on CI) (#31) Signed-off-by: AlexDBlack <blacka101@gmail.com> * [WIP] More CUDA stuff (#26) * initial commit Signed-off-by: raver119 <raver119@gmail.com> * LRN BP CUDA Signed-off-by: raver119 <raver119@gmail.com> * less memory Signed-off-by: raver119 <raver119@gmail.com> * Fixed bug with crop_and_resize op helper. * get rid of unnecessary index-calculation dunction Signed-off-by: Yurii <yurii@skymind.io> * Fixed sort with nth_element cuda-based helper. * Refactored nth_element. * Refactored nth_element op and tests. * Modified usage of dim array with sortTad routine. * Refactored main routine of helper for non_max_image_suppression op. * non_max_image_suppression op helper with cuda kernel implementation. Initial revision. * fix vol2col cuda kernel * meh Signed-off-by: raver119 <raver119@gmail.com> * topK concept Signed-off-by: raver119 <raver119@gmail.com> * unsorted topK with scanWitdh of 1 Signed-off-by: raver119 <raver119@gmail.com> * correct vol2col tests * sorted/unsorted topK Signed-off-by: raver119 <raver119@gmail.com> * implementation and fixing col2im/col2vol * Corrected usage flags with input/output with reverse op. * dup is const now Signed-off-by: raver119 <raver119@gmail.com> * percentile op Signed-off-by: raver119 <raver119@gmail.com> * group tests for mapool2d Signed-off-by: Yurii <yurii@skymind.io> * special test for george Signed-off-by: raver119 <raver119@gmail.com> * less threads for sortTad Signed-off-by: raver119 <raver119@gmail.com> * provide conv2d for cuda Signed-off-by: Yurii <yurii@skymind.io> * remove auther in sort tad kernel code Signed-off-by: Yurii <yurii@skymind.io> * provide depthwise_conv2d for cuda Signed-off-by: Yurii <yurii@skymind.io> * - max_pooling_with_argmax - null check for special use Signed-off-by: raver119 <raver119@gmail.com> * dts cuda Signed-off-by: raver119 <raver119@gmail.com> * provide sconv2d for cuda Signed-off-by: Yurii <yurii@skymind.io> * std cuda Signed-off-by: raver119 <raver119@gmail.com> * Refactored non_max_suppression op to conform TF implementation. * Improved suppression helper. * provide pooling3d for cuda Signed-off-by: Yurii <yurii@skymind.io> * minor lstm rearrangements Signed-off-by: raver119 <raver119@gmail.com> * more of minor lstm rearrangements Signed-off-by: raver119 <raver119@gmail.com> * (bi)dynamic_rnn Signed-off-by: raver119 <raver119@gmail.com> * templates init order Signed-off-by: raver119 <raver119@gmail.com> * Refactored non_max_suppression op. * Added cuda kernel for non_max_suppression. * CPU sort by key/value Signed-off-by: raver119 <raver119@gmail.com> * CPU sort TAD by key/value Signed-off-by: raver119 <raver119@gmail.com> * CPU sort TAD by key/value tests Signed-off-by: raver119 <raver119@gmail.com> * Eliminate compiler error with cuda implementation. * - repaired gradCheck in cuda - provide conv2d_bp for cuda Signed-off-by: Yurii <yurii@skymind.io> * missed signature Signed-off-by: raver119 <raver119@gmail.com> * provide depthwise_conv2d_bp for cuda Signed-off-by: Yurii <yurii@skymind.io> * Implementation of lup helper with cuda kernel. Initial commit. * further work on backprops for convolutions Signed-off-by: Yurii <yurii@skymind.io> * CUDA linear sort by key/val Signed-off-by: raver119 <raver119@gmail.com> * CUDA tad sort by key/val Signed-off-by: raver119 <raver119@gmail.com> * start providing of backprop for pooling2d/3d Signed-off-by: Yurii <yurii@skymind.io> * Added atomicAdd for bool datatype. * dynamic partition concept Signed-off-by: raver119 <raver119@gmail.com> * dynamic partition concept Signed-off-by: raver119 <raver119@gmail.com> * dynamic partition scalar CUDA Signed-off-by: raver119 <raver119@gmail.com> * important comment Signed-off-by: raver119 <raver119@gmail.com> * fix pooling2d/3d backprop helpers Signed-off-by: Yurii <yurii@skymind.io> * Added non-linear test with dynamic_partition. * Improved test for dynamic_partition. * dynamic_partition TAD concept Signed-off-by: raver119 <raver119@gmail.com> * - dynamic_partition TAD CUDA impl - dynamic_partition TAD CPU fix Signed-off-by: raver119 <raver119@gmail.com> * - rewrite cpu code for usampling2d/3d - write cuda code for usampling2d/3d Signed-off-by: Yurii <yurii@skymind.io> * dynamic_stitch CUDA vector case Signed-off-by: raver119 <raver119@gmail.com> * dynamic_stitch CUDA TAD case concept Signed-off-by: raver119 <raver119@gmail.com> * dynamic_stitch CUDA TAD case impl Signed-off-by: raver119 <raver119@gmail.com> * Added tests for dynamic_stitch 3D-4D cases. * minor tests tweaks Signed-off-by: raver119 <raver119@gmail.com> * Fixed type check for dynamic stitch. * min/max bp Signed-off-by: raver119 <raver119@gmail.com> * rewrite code for upsampling2d/3d cpu Signed-off-by: Yurii <yurii@skymind.io> * reduce min/max/norm_max bp Signed-off-by: raver119 <raver119@gmail.com> * lup implementation. Additional enhancements. * provide code for upsamling2d/3d backprop Signed-off-by: Yurii <yurii@skymind.io> * weightedCrossEntropyWithLogits Signed-off-by: raver119 <raver119@gmail.com> * Fixed template math atomicMul for 64bit ints. * Refactored dynamic_partition_bp op. * inverseBroadcast fix Signed-off-by: raver119 <raver119@gmail.com> * DynamicPartitionBP test datatype fixed. * - nd4j_atomicMul Windows fix - cpu/NDArrayLambda.hpp excluded from CUDA Signed-off-by: raver119 <raver119@gmail.com>
2019-06-27 17:37:04 +02:00
template NDArray NDArrayFactory::create(const std::vector<uint16_t> &values, nd4j::LaunchContext * context);
2019-06-06 14:21:15 +02:00
template NDArray NDArrayFactory::create(const std::vector<int8_t> &values, nd4j::LaunchContext * context);
template NDArray NDArrayFactory::create(const std::vector<uint8_t> &values, nd4j::LaunchContext * context);
template NDArray NDArrayFactory::create(const std::vector<bool> &values, nd4j::LaunchContext * context);
////////////////////////////////////////////////////////////////////////
template <typename T>
NDArray* NDArrayFactory::empty_(nd4j::LaunchContext * context) {
auto shapeInfo = ShapeBuilders::createScalarShapeInfo(DataTypeUtils::fromT<T>(), context->getWorkspace());
ArrayOptions::setPropertyBit(shapeInfo, ARRAY_EMPTY);
auto result = new NDArray(nullptr, shapeInfo, context, false);
RELEASE(shapeInfo, context->getWorkspace());
return result;
}
BUILD_SINGLE_TEMPLATE(template NDArray* NDArrayFactory::empty_, (nd4j::LaunchContext * context), LIBND4J_TYPES);
NDArray* NDArrayFactory::empty_(nd4j::DataType dataType, nd4j::LaunchContext * context) {
if (context == nullptr)
context = nd4j::LaunchContext ::defaultContext();
auto shapeInfo = ShapeBuilders::createScalarShapeInfo(dataType, context->getWorkspace());
ArrayOptions::setPropertyBit(shapeInfo, ARRAY_EMPTY);
auto result = new NDArray(nullptr, shapeInfo, context, false);
RELEASE(shapeInfo, context->getWorkspace());
return result;
}
////////////////////////////////////////////////////////////////////////
template <typename T>
NDArray NDArrayFactory::empty(nd4j::LaunchContext * context) {
return empty(DataTypeUtils::fromT<T>(), context);
}
BUILD_SINGLE_TEMPLATE(template NDArray NDArrayFactory::empty, (nd4j::LaunchContext * context), LIBND4J_TYPES);
////////////////////////////////////////////////////////////////////////
NDArray NDArrayFactory::empty(nd4j::DataType dataType, nd4j::LaunchContext * context) {
auto shapeInfo = ShapeBuilders::createScalarShapeInfo(dataType, context->getWorkspace());
ArrayOptions::setPropertyBit(shapeInfo, ARRAY_EMPTY);
NDArray result(nullptr, shapeInfo, context, false);
RELEASE(shapeInfo, context->getWorkspace());
return result;
}
////////////////////////////////////////////////////////////////////////
NDArray* NDArrayFactory::valueOf(const std::vector<Nd4jLong>& shape, const NDArray& value, const char order, nd4j::LaunchContext * context) {
auto res = NDArrayFactory::create_(order, shape, value.dataType(), context);
res->assign(const_cast<NDArray&>(value));
return res;
}
////////////////////////////////////////////////////////////////////////
NDArray* NDArrayFactory::create_( const char order, const std::vector<Nd4jLong> &shape, nd4j::DataType dataType, nd4j::LaunchContext * context) {
return new NDArray(order, shape, dataType, context);
}
////////////////////////////////////////////////////////////////////////
template <typename T>
NDArray NDArrayFactory::create(T* buffer, const char order, const std::initializer_list<Nd4jLong>& shape, nd4j::LaunchContext * context) {
if ((int) shape.size() > MAX_RANK)
throw std::invalid_argument("NDArrayFactory::create: Rank of NDArray can't exceed 32");
std::vector<Nd4jLong> shp(shape);
ShapeDescriptor descriptor(DataTypeUtils::fromT<T>(), order, shp);
std::shared_ptr<DataBuffer> pBuffer = std::make_shared<DataBuffer>(buffer, descriptor.arrLength() * sizeof(T), descriptor.dataType(), false, context->getWorkspace());
NDArray result(pBuffer, descriptor, context);
return result;
}
template NDArray NDArrayFactory::create(double* buffer, const char order, const std::initializer_list<Nd4jLong>& shape, nd4j::LaunchContext * context);
template NDArray NDArrayFactory::create(float* buffer, const char order, const std::initializer_list<Nd4jLong>& shape, nd4j::LaunchContext * context);
template NDArray NDArrayFactory::create(float16* buffer, const char order, const std::initializer_list<Nd4jLong>& shape, nd4j::LaunchContext * context);
template NDArray NDArrayFactory::create(bfloat16* buffer, const char order, const std::initializer_list<Nd4jLong>& shape, nd4j::LaunchContext * context);
template NDArray NDArrayFactory::create(Nd4jLong * buffer, const char order, const std::initializer_list<Nd4jLong>& shape, nd4j::LaunchContext * context);
template NDArray NDArrayFactory::create(int* buffer, const char order, const std::initializer_list<Nd4jLong>& shape, nd4j::LaunchContext * context);
template NDArray NDArrayFactory::create(bool* buffer, const char order, const std::initializer_list<Nd4jLong>& shape, nd4j::LaunchContext * context);
template NDArray NDArrayFactory::create(uint8_t * buffer, const char order, const std::initializer_list<Nd4jLong>& shape, nd4j::LaunchContext * context);
template NDArray NDArrayFactory::create(int8_t* buffer, const char order, const std::initializer_list<Nd4jLong>& shape, nd4j::LaunchContext * context);
template NDArray NDArrayFactory::create(int16_t* buffer, const char order, const std::initializer_list<Nd4jLong>& shape, nd4j::LaunchContext * context);
NDArray NDArrayFactory::string(char order, const std::vector<Nd4jLong> &shape, const std::initializer_list<const char *> &strings, nd4j::LaunchContext * context) {
std::vector<const char*> vec(strings);
return NDArrayFactory::string(order, shape, vec, context);
}
NDArray NDArrayFactory::string(char order, const std::vector<Nd4jLong> &shape, const std::vector<const char *> &strings, nd4j::LaunchContext * context) {
std::vector<std::string> vec(strings.size());
int cnt = 0;
for (auto s:strings)
vec[cnt++] = std::string(s);
return NDArrayFactory::string(order, shape, vec, context);
}
NDArray NDArrayFactory::string(char order, const std::vector<Nd4jLong> &shape, const std::initializer_list<std::string> &string, nd4j::LaunchContext * context) {
std::vector<std::string> vec(string);
return NDArrayFactory::string(order, shape, vec, context);
}
NDArray* NDArrayFactory::string_(char order, const std::vector<Nd4jLong> &shape, const std::initializer_list<const char *> &strings, nd4j::LaunchContext * context) {
std::vector<const char*> vec(strings);
return NDArrayFactory::string_(order, shape, vec, context);
}
NDArray* NDArrayFactory::string_(char order, const std::vector<Nd4jLong> &shape, const std::vector<const char *> &strings, nd4j::LaunchContext * context) {
std::vector<std::string> vec(strings.size());
int cnt = 0;
for (auto s:strings)
vec[cnt++] = std::string(s);
return NDArrayFactory::string_(order, shape, vec, context);
}
NDArray* NDArrayFactory::string_(char order, const std::vector<Nd4jLong> &shape, const std::initializer_list<std::string> &string, nd4j::LaunchContext * context) {
std::vector<std::string> vec(string);
return NDArrayFactory::string_(order, shape, vec, context);
}
NDArray NDArrayFactory::string(char order, const std::vector<Nd4jLong> &shape, const std::vector<std::string> &string, nd4j::LaunchContext * context) {
if (context == nullptr)
context = nd4j::LaunchContext ::defaultContext();
auto headerLength = ShapeUtils::stringBufferHeaderRequirements(string.size());
std::vector<Nd4jLong> offsets(string.size() + 1);
Nd4jLong dataLength = 0;
for (int e = 0; e < string.size(); e++) {
offsets[e] = dataLength;
dataLength += string[e].length();
}
offsets[string.size()] = dataLength;
std::shared_ptr<DataBuffer> pBuffer = std::make_shared<DataBuffer>(headerLength + dataLength, DataType::UTF8, context->getWorkspace(), true);
NDArray res(pBuffer, ShapeDescriptor(DataType::UTF8, order, shape), context);
res.setAttached(context->getWorkspace() != nullptr);
if (res.lengthOf() != string.size())
throw std::invalid_argument("Number of strings should match length of array");
memcpy(res.buffer(), offsets.data(), offsets.size() * sizeof(Nd4jLong));
auto data = static_cast<int8_t*>(res.buffer()) + headerLength;
int resLen = res.lengthOf();
for (int e = 0; e < resLen; e++) {
auto length = offsets[e+1] - offsets[e];
auto cdata = data + offsets[e];
memcpy(cdata, string[e].c_str(), string[e].length());
}
res.tickWriteHost();
res.syncToDevice();
return res;
}
NDArray* NDArrayFactory::string_(char order, const std::vector<Nd4jLong> &shape, const std::vector<std::string> &string, nd4j::LaunchContext * context) {
auto res = new NDArray();
*res = NDArrayFactory::string(order, shape, string, context);
return res;
}
}