Yurii Shyrma fe47f52896
Oleh tenzor mmul (#231)
* Libnd4j: TensorMMul backprop op #8174, raw implementation

Signed-off-by: Oleg <oleg.semeniv@gmail.com>

* Libnd4j: TensorMMul backprop op #8174 merge master and some corrections

Signed-off-by: Oleg <oleg.semeniv@gmail.com>

* Libnd4j: TensorMMul backprop op #8174 algorithm update, need testing, sync with  master

* Libnd4j: TensorMMul backprop op #8174 fixed incorrect B axes calculation

Signed-off-by: Oleg <oleg.semeniv@gmail.com>

* Libnd4j: TensorMMul backprop op #8174 optimize axes identification and fix bug of indeces overlapping, added first test. need testing with different shapes

Signed-off-by: Oleg <oleg.semeniv@gmail.com>

* Libnd4j: TensorMMul backprop op #8174 some fixes and improvements need more testing

Signed-off-by: Oleg <oleg.semeniv@gmail.com>

* Libnd4j: TensorMMul backprop op #8174 fixed order of matrix multiply

Signed-off-by: Oleg <oleg.semeniv@gmail.com>

* Libnd4j: TensorMMul backprop op #8174 fixed issue of incorrect axes definition, add tests based on TF, need additional testing for case dLdC not equal 1

Signed-off-by: Oleg <oleg.semeniv@gmail.com>

* Libnd4j: TensorMMul backprop op #8174 fixed scalar case add test

Signed-off-by: Oleg <oleg.semeniv@gmail.com>

* Libnd4j: TensorMMul backprop op #8174 fixed bp algorithm, axes definition, need some mode testing with different orders combination f,c; c,f f,f and add some checks for inputs

Signed-off-by: Oleg <oleg.semeniv@gmail.com>

* Libnd4j: TensorMMul backprop op #8174 some checks and corrections added tests, exists the problem with different input orders support A-f B-c and A-f B-f

Signed-off-by: Oleg <oleg.semeniv@gmail.com>

* Libnd4j: TensorMMul backprop op #8174 sync master

Signed-off-by: Oleg <oleg.semeniv@gmail.com>

* - correct bug in MmulHelper::tensorDot(a, b, c, axes_a, axes_b,permutForC)

Signed-off-by: Yurii <iuriish@yahoo.com>

* Libnd4j: TensorMMul backprop op #8174 code clean up and refactoring

Signed-off-by: Oleg <oleg.semeniv@gmail.com>

* - add check for linspase ordered permutations in ShapeUtils::evalShapeForTensorDot

Signed-off-by: Yurii <iuriish@yahoo.com>

* - provide additional code in shape::reshape stuff in order to reduce amount of allocation/copy operations during reshaping procedure

Signed-off-by: Yurii <iuriish@yahoo.com>

* - further work on problem of wrong shape evaluation during permute/reshape procedures

Signed-off-by: Yurii <iuriish@yahoo.com>

* - still looking for bug reason in reshape/permute stuff

Signed-off-by: Yurii <iuriish@yahoo.com>

* - correct bug in transform cuda native ops

Signed-off-by: Yurii <iuriish@yahoo.com>

* - correct bug in NDArray::assign

Signed-off-by: Yurii <iuriish@yahoo.com>

* - remove old shape::reshape stuff

Signed-off-by: Yurii <iuriish@yahoo.com>

* - add possibility to disable copy of old buffer to new buffer during reshape operation in NDArray class

Signed-off-by: Yurii <iuriish@yahoo.com>

* - correct bug in tensorDot which had to do with wrong pointers assigments

Signed-off-by: Yurii <iuriish@yahoo.com>

Co-authored-by: Oleh <oleg.semeniv@gmail.com>
2020-02-13 20:33:54 +03:00

154 lines
5.1 KiB
C++

/*******************************************************************************
* 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
******************************************************************************/
//
// @author raver119@gmail.com
//
#include <op_boilerplate.h>
#if NOT_EXCLUDED(OP_squeeze)
#include <ops/declarable/CustomOperations.h>
namespace nd4j {
namespace ops {
CUSTOM_OP_IMPL(squeeze, 1, 1, true, 0, -2) {
auto input = INPUT_VARIABLE(0);
auto output = OUTPUT_VARIABLE(0);
std::vector<int> axis;
if (block.numI() > 0)
for (int e = 0; e < block.numI(); e++) {
int _a = INT_ARG(e);
if (_a < 0)
_a += input->rankOf();
axis.emplace_back(_a);
}
else if (block.width() > 1) {
auto a = INPUT_VARIABLE(1);
for (Nd4jLong e = 0; e < a->lengthOf(); e++) {
int _a = a->e<int>(e);
if (_a < 0)
_a += input->rankOf();
axis.emplace_back(_a);
}
}
if (input->rankOf() == 0 || (input->rankOf() == 1 && input->lengthOf() == 1)) {
output->assign(input);
return Status::OK();
}
std::vector<Nd4jLong> shape;
if (axis.size() == 0) {
for (int d = 0; d < input->rankOf(); d++)
if (input->sizeAt(d) > 1)
shape.emplace_back(input->sizeAt(d));
} else {
for (int d = 0; d < input->rankOf(); d++) {
if (input->sizeAt(d) == 1) {
if (std::find(axis.begin(), axis.end(), d) == axis.end())
shape.emplace_back(input->sizeAt(d));
} else shape.emplace_back(input->sizeAt(d));
}
}
if (block.isInplace()) {
output->reshapei(input->ordering(), shape, false);
} else {
auto tmp = input->reshape(input->ordering(), shape);
output->assign(tmp);
}
return Status::OK();
}
DECLARE_TYPES(squeeze) {
getOpDescriptor()
->setAllowedInputTypes(nd4j::DataType::ANY)
->setSameMode(true);
}
DECLARE_SHAPE_FN(squeeze) {
auto shapeList = SHAPELIST();
Nd4jLong* newShape;
auto in = inputShape->at(0);
auto rank = shape::rank(in);
auto length = shape::length(in);
if (rank == 0 || (rank == 1 && length == 1)) {
shapeList->push_back(ConstantShapeHelper::getInstance()->scalarShapeInfo(ArrayOptions::dataType(in)));
return shapeList;
}
std::vector<int> axis;
if (block.numI() > 0)
for (int e = 0; e < block.numI(); e++) {
int _a = INT_ARG(e);
if (_a < 0)
_a += rank;
axis.emplace_back(_a);
}
else if (block.width() > 1) {
auto a = INPUT_VARIABLE(1);
for (int e = 0; e < a->lengthOf(); e++) {
int _a = a->e<int>(e);
if (_a < 0)
_a += rank;
axis.emplace_back(_a);
}
}
auto order = shape::order(in);
auto oldShape = shape::shapeOf(in);
std::vector<Nd4jLong> shape;
if (axis.size() == 0) {
for (int d = 0; d < rank; d++)
if (oldShape[d] > 1)
shape.emplace_back(oldShape[d]);
} else {
for (int d = 0; d < rank; d++) {
if (oldShape[d] == 1) {
if (std::find(axis.begin(), axis.end(), d) == axis.end())
shape.emplace_back(oldShape[d]);
} else shape.emplace_back(oldShape[d]);
}
}
if (shape.size() == 0) {
shapeList->push_back(ConstantShapeHelper::getInstance()->scalarShapeInfo(ArrayOptions::dataType(in)));
return shapeList;
}
newShape = ConstantShapeHelper::getInstance()->createShapeInfo(ArrayOptions::dataType(in), order, shape);
shapeList->push_back(newShape);
return shapeList;
}
}
}
#endif