109 lines
5.0 KiB
C++
109 lines
5.0 KiB
C++
/*******************************************************************************
|
|
* Copyright (c) 2019-2020 Konduit K.K.
|
|
*
|
|
* 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 Oleh Semeniv (oleg.semeniv@gmail.com)
|
|
//
|
|
|
|
#include <ops/declarable/helpers/updatersHelpers.h>
|
|
#include <execution/Threads.h>
|
|
#include <math/platformmath.h>
|
|
#include <math/templatemath.h>
|
|
|
|
namespace sd {
|
|
namespace ops {
|
|
namespace helpers {
|
|
|
|
//////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////
|
|
template <typename T>
|
|
static void adaDeltaUpdater_(const NDArray& gradient, const NDArray& initStateMsg, const NDArray& initStateMsdx,
|
|
NDArray& update, NDArray& stateMsg, NDArray& stateMsdx, const double dRho, const double dEpsilon) {
|
|
|
|
const T* grad = gradient.bufferAsT<T>();
|
|
const T* initMsg = initStateMsg.bufferAsT<T>();
|
|
const T* initMsdx = initStateMsdx.bufferAsT<T>();
|
|
|
|
T* up = update.bufferAsT<T>();
|
|
T* stMsg = stateMsg.bufferAsT<T>();
|
|
T* stMsdx = stateMsdx.bufferAsT<T>();
|
|
|
|
const T rho = static_cast<T>(dRho);
|
|
const T epsilon = static_cast<T>(dEpsilon);
|
|
const T rhoT = (1 - rho);
|
|
|
|
bool bEws1 = 1 == gradient.ews() && 1 == update.ews() && 1 == stateMsg.ews() && 1 == initStateMsg.ews() && 1 == stateMsdx.ews() && 1 == initStateMsdx.ews();
|
|
bool bSameOrdering = gradient.ordering() == update.ordering() &&
|
|
update.ordering() == stateMsdx.ordering() &&
|
|
stateMsdx.ordering() == initStateMsdx.ordering() &&
|
|
stateMsdx.ordering() == initStateMsg.ordering() && stateMsg.ordering() == initStateMsg.ordering();
|
|
|
|
if (bEws1 && bSameOrdering) {
|
|
|
|
auto func = PRAGMA_THREADS_FOR{
|
|
for (auto i = start; i < stop; i++) {
|
|
stMsg[i] = rho * initMsg[i] + grad[i] * grad[i] * rhoT;
|
|
|
|
up[i] = grad[i] * (sd::math::nd4j_sqrt<T, T>(initMsdx[i] + epsilon) / sd::math::nd4j_sqrt<T, T>(stMsg[i] + epsilon));
|
|
|
|
stMsdx[i] = rho * initMsdx[i] + up[i] * up[i] * rhoT;
|
|
}
|
|
};
|
|
|
|
samediff::Threads::parallel_for(func, 0, gradient.lengthOf(), 1);
|
|
return;
|
|
}
|
|
|
|
|
|
bool bXZsame = shape::haveSameShapeAndStrides(gradient.shapeInfo(), update.shapeInfo());
|
|
bool bXInMsgSame = shape::haveSameShapeAndStrides(gradient.shapeInfo(), initStateMsg.shapeInfo());
|
|
bool bXStMsgSame = shape::haveSameShapeAndStrides(gradient.shapeInfo(), stateMsg.shapeInfo());
|
|
bool bXInMsdxSame = shape::haveSameShapeAndStrides(gradient.shapeInfo(), initStateMsdx.shapeInfo());
|
|
bool bXStMsdxSame = shape::haveSameShapeAndStrides(gradient.shapeInfo(), stateMsdx.shapeInfo());
|
|
|
|
auto func = PRAGMA_THREADS_FOR{
|
|
|
|
int coords[MAX_RANK];
|
|
for (auto i = start; i < gradient.lengthOf(); i++) {
|
|
shape::index2coordsCPU(start, i, gradient.shapeInfo(), coords);
|
|
const auto xOffset = shape::getOffset(gradient.shapeInfo(), coords);
|
|
const auto zOffset = bXZsame ? xOffset : shape::getOffset(update.shapeInfo(), coords);
|
|
const auto initMsgOffset = bXInMsgSame ? xOffset : shape::getOffset(initStateMsg.shapeInfo(), coords);
|
|
const auto stMsgOffset = bXStMsgSame ? xOffset : shape::getOffset(stateMsg.shapeInfo(), coords);
|
|
const auto initMsdxOffset = bXInMsdxSame ? xOffset : shape::getOffset(initStateMsdx.shapeInfo(), coords);
|
|
const auto stMsdxOffset = bXStMsdxSame ? xOffset : shape::getOffset(stateMsdx.shapeInfo(), coords);
|
|
|
|
|
|
stMsg[stMsgOffset] = rho * initMsg[initMsgOffset] + grad[xOffset] * grad[xOffset] * rhoT;
|
|
|
|
up[zOffset] = grad[xOffset] * (sd::math::nd4j_sqrt<T, T>(initMsdx[initMsdxOffset] + epsilon) / sd::math::nd4j_sqrt<T, T>(stMsg[stMsgOffset] + epsilon));
|
|
|
|
stMsdx[stMsdxOffset] = rho * initMsdx[initMsdxOffset] + up[zOffset] * up[zOffset] * rhoT;
|
|
}
|
|
};
|
|
|
|
samediff::Threads::parallel_for(func, 0, gradient.lengthOf(), 1);
|
|
return;
|
|
}
|
|
|
|
void updaterAdaDelta(sd::LaunchContext* context, const NDArray& gradient, const NDArray& initStateMsg, const NDArray& initStateMsdx,
|
|
NDArray& update, NDArray& stateMsg, NDArray& stateMsdx, const double dRho, const double dEpsilon) {
|
|
BUILD_SINGLE_SELECTOR(gradient.dataType(), adaDeltaUpdater_, (gradient, initStateMsg, initStateMsdx, update, stateMsg, stateMsdx, dRho, dEpsilon), FLOAT_TYPES);
|
|
}
|
|
|
|
}
|
|
}
|
|
}
|