/******************************************************************************* * 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 Yurii Shyrma (iuriish@yahoo.com), created on 22.11.2017 // #include #if NOT_EXCLUDED(OP_cosine_distance_loss) #include #include namespace nd4j { namespace ops { ////////////////////////////////////////////////////////////////////////// CUSTOM_OP_IMPL(cosine_distance_loss, 3, 1, false, 0, 2) { auto predictions = INPUT_VARIABLE(0); auto weights = INPUT_VARIABLE(1); auto labels = INPUT_VARIABLE(2); auto output = OUTPUT_VARIABLE(0); int reductionMode = INT_ARG(0); // 0 - "none"; 1 - "weighted_sum"; 2 - "weighted_mean"; 3 - "weighted_sum_by_nonzero_weights" int dim = INT_ARG(1); // axis along which sum will be made if(dim < 0) dim += labels->rankOf(); // labels and predictions must have the same shapes REQUIRE_TRUE(labels->isSameShape(predictions), 0, "COSINE_DISTANCE_LOSS OP: labels and predictions arrays must have the same shapes, but got %s and %s correspondingly !", ShapeUtils::shapeAsString(labels).c_str(), ShapeUtils::shapeAsString(predictions).c_str()); // regard 4 possible reduction modes below REQUIRE_TRUE(reductionMode==0 || reductionMode==1 || reductionMode==2 || reductionMode==3, 0, "COSINE_DISTANCE_LOSS OP: reduction mode value is not acceptable, possible values are 0, 1, 2, 3, but got %i instead!", reductionMode); // input dimension can't be larger than labels/predictions/weights rank REQUIRE_TRUE(dim < labels->rankOf(), 0, "COSINE_DISTANCE_LOSS OP: input reduction dimension (got %i) must be < labels rank %i!", dim, labels->rankOf()); if(!output->isScalar()) { // weights array can be single scalar or has the same shape as output, and must be broadcastable to output shape REQUIRE_TRUE(weights->isScalar() || weights->rankOf() == output->rankOf(), 0, "SOFTMAX_CROSS_ENTROPY_LOSS OP: weights array should be scalar or have the same rank as output array, but got %i and %i correspondingly!", weights->rankOf(), output->rankOf()); // check whether broadcast operation is possible for weights array REQUIRE_TRUE(weights->isScalar() || ShapeUtils::areShapesBroadcastable(*weights, *output), 0, "COSINE_DISTANCE_LOSS OP: shapes of weights and output arrays should be broadcastable, but got weights = %s and output = %s instead!", ShapeUtils::shapeAsString(weights).c_str(), ShapeUtils::shapeAsString(labels).c_str()); } NDArray E = 1. - (*predictions * *labels).reduceAlongDims(reduce::Sum, {dim}, true); // perform weights broadcasting/tile to E if it is necessary auto weightsBroad = weights; if(!weights->isScalar() && !weights->isSameShape(&E)) weightsBroad = new NDArray(weights->tileToShape(E.getShapeInfo())); // multiply E on weights E *= (*weightsBroad); switch (reductionMode) { case 0: // 0 - "none", un-reduced weighted losses with the same shape as labels. output->assign(&E); break; case 1: { // 1 - "weighted_sum", output is scalar and equal to sum of all elements of E array output->assign(E.reduceNumber(reduce::Sum)); break; } case 2: { // 2 - "weighted_mean", output is scalar and equal to sum of all elements of E array divided by sum of all elements of weightsBroad array NDArray sum; if (weights->isScalar()) sum = *weights * E.lengthOf(); else sum = weightsBroad->reduceNumber(reduce::Sum); if (sum.e(0) == 0.) *output = 0.; else output->assign(E.reduceNumber(reduce::Sum) / sum); break; } case 3: { // 3 - "weighted_sum_by_nonzero_weights", output is scalar and equal to scalar sum of all elements of E array divided by number of non-zero weights Nd4jLong numOfNonZeroWeights = 0; if(weights->isScalar()) { if(weights->e(0) != 0.) numOfNonZeroWeights = E.lengthOf(); } else numOfNonZeroWeights = E.reduceNumber(reduce::CountNonZero).e(0); if (numOfNonZeroWeights == 0) *output = 0.; else output->assign(E.reduceNumber(reduce::Sum) / double(numOfNonZeroWeights)); break; } } STORE_RESULT(*output); if(weightsBroad != weights) delete weightsBroad; return Status::OK(); } ////////////////////////////////////////////////////////////////////////// DECLARE_TYPES(cosine_distance_loss) { getOpDescriptor()->setAllowedInputTypes(nd4j::DataType::ANY)->setAllowedOutputTypes({ALL_FLOATS}); } ////////////////////////////////////////////////////////////////////////// DECLARE_SHAPE_FN(cosine_distance_loss) { // labels and predictions must have the same shapes auto predictionsShapeInfo = inputShape->at(0); auto weightsShapeInfo = inputShape->at(1); auto labelsShapeInfo = inputShape->at(2); int dim = INT_ARG(1); if(dim < 0) dim += labelsShapeInfo[0]; // labels and predictions must have the same shapes REQUIRE_TRUE(shape::shapeEquals(labelsShapeInfo, predictionsShapeInfo), 0, "COSINE_DISTANCE_LOSS OP: labels and predictions arrays must have the same shapes, but got %s and %s correspondingly !", ShapeUtils::shapeAsString(labelsShapeInfo).c_str(), ShapeUtils::shapeAsString(predictionsShapeInfo).c_str()); // input dimension can't be larger than labels/predictions/weights rank REQUIRE_TRUE(dim < labelsShapeInfo[0], 0, "COSINE_DISTANCE_LOSS OP: input reduction dimension (got %i) must be < labels rank %i!", dim, labelsShapeInfo[0]); DataType outType = DataTypeUtils::pickFloatingType(ArrayOptions::dataType(predictionsShapeInfo)); // evaluate output shapeInfo Nd4jLong* outShapeInfo = nullptr; if(INT_ARG(0) != 0) // in this case output is scalar outShapeInfo = ConstantShapeHelper::getInstance()->scalarShapeInfo(outType); else { // in this case output has the same shape as labels reduced by dim axis std::vector dimensions = {dim}; outShapeInfo = ShapeUtils::evalReduceShapeInfo(shape::order(predictionsShapeInfo), dimensions, predictionsShapeInfo, outType, true, false, block.getWorkspace()); // weights array can be single scalar or has the same rank as output, and must be broadcastable to output REQUIRE_TRUE(shape::isScalar(weightsShapeInfo) || shape::rank(weightsShapeInfo) == shape::rank(outShapeInfo), 0, "COSINE_DISTANCE_LOSS OP: weights array should be scalar or have the same rank as output array, but got %i and %i correspondingly!", shape::rank(weightsShapeInfo), shape::rank(outShapeInfo)); // check whether broadcast operation is possible for weights array REQUIRE_TRUE(shape::isScalar(weightsShapeInfo) || ShapeUtils::areShapesBroadcastable(weightsShapeInfo, outShapeInfo), 0, "COSINE_DISTANCE_LOSS OP: shapes of weights and output arrays should be broadcastable, but got weights = %s and output = %s instead!", ShapeUtils::shapeAsString(weightsShapeInfo).c_str(), ShapeUtils::shapeAsString(outShapeInfo).c_str()); } return SHAPELIST(outShapeInfo); } ////////////////////////////////////////////////////////////////////////// CUSTOM_OP_IMPL(cosine_distance_loss_grad, 3, 3, false, 0, 2) { auto predictions = INPUT_VARIABLE(0); auto weights = INPUT_VARIABLE(1); auto labels = INPUT_VARIABLE(2); auto dLdp = OUTPUT_VARIABLE(0); // dL/dpredictions auto dLdw = OUTPUT_VARIABLE(1); // dL/dweights auto dLdl = OUTPUT_VARIABLE(2); // dL/dlabels int reductionMode = INT_ARG(0); // 0 - "none"; 1 - "weighted_sum"; 2 - "weighted_mean"; 3 - "weighted_sum_by_nonzero_weights" // take into account Alex's proposition to treat "none" the same as "weighted_sum" mode when calculating gradients if(reductionMode == 0) reductionMode = 1; int dim = INT_ARG(1); // axis along which sum will be made if(dim < 0) dim += labels->rankOf(); std::vector dimensions = {dim}; // input validation REQUIRE_TRUE(labels->isSameShape(predictions), 0, "COSINE_DISTANCE_LOSS_GRAD OP: labels and predictions arrays must have the same shapes, but got %s and %s correspondingly !", ShapeUtils::shapeAsString(labels).c_str(), ShapeUtils::shapeAsString(predictions).c_str()); // only 4 possible reduction modes exist REQUIRE_TRUE(reductionMode==0 || reductionMode==1 || reductionMode==2 || reductionMode==3, 0, "COSINE_DISTANCE_LOSS_GRAD OP: reduction mode value is not acceptable, possible values are 0, 1, 2, 3, but got %i instead!", reductionMode); auto lossShapeInfo = ShapeUtils::evalReduceShapeInfo(predictions->ordering(), dimensions, predictions->getShapeInfo(), true, false, block.getWorkspace()); // weights array can be single scalar or has the same shape as loss, and must be broadcastable to loss shape REQUIRE_TRUE(weights->isScalar() || weights->rankOf() == shape::rank(lossShapeInfo), 0, "COSINE_DISTANCE_LOSS_GRAD OP: weights array should be scalar or have the same rank as loss array, but got %i and %i correspondingly!", weights->rankOf(), shape::rank(lossShapeInfo)); // check whether broadcast operation is possible for weights array REQUIRE_TRUE(weights->isScalar() || ShapeUtils::areShapesBroadcastable(weights->getShapeInfo(), lossShapeInfo), 0, "COSINE_DISTANCE_LOSS_GRAD OP: shapes of weights and loss arrays should be broadcastable, but got weights = %s and loss = %s instead!", ShapeUtils::shapeAsString(weights).c_str(), ShapeUtils::shapeAsString(lossShapeInfo).c_str()); // input dimension can't be larger than labels/predictions/weights rank REQUIRE_TRUE(dim < labels->rankOf(), 0, "COSINE_DISTANCE_LOSS_GRAD OP: input reduction dimension (got %i) must be < labels rank %i!", dim, labels->rankOf()); NDArray E = 1. - (*predictions * *labels).reduceAlongDims(reduce::Sum, {dim}, true); // perform weights broadcasting/tile to E if it is necessary auto weightsBroad = weights; if(!weights->isScalar() && !weights->isSameShape(&E)) weightsBroad = new NDArray(weights->tileToShape(E.getShapeInfo())); dLdp->assign(-*labels); dLdl->assign(-*predictions); switch (reductionMode) { case 1: { // 1 - "none" and "weighted_sum", output is scalar and equal to sum of all elements of E array *dLdp *= *weightsBroad; *dLdl *= *weightsBroad; if(weights->isScalar() || weights->lengthOf() == 1) { dLdw->assign(E.reduceNumber(reduce::Sum)); } else { if(weights != weightsBroad) { std::vector axesToReduceAlong = ShapeUtils::evalBroadcastBackwardAxis(weights->getShapeInfo(), weightsBroad->getShapeInfo()); E.reduceAlongDimension(reduce::Sum, dLdw, axesToReduceAlong, true, false, false); } else dLdw->assign(E); } break; } case 2: { // 2 - "weighted_mean", output is scalar and equal to sum of all elements of E array divided by sum of all elements of weightsBroad array NDArray sum; if (weights->isScalar()) sum = (*weights) * E.lengthOf(); else sum = weightsBroad->reduceNumber(reduce::Sum); if (sum.e(0) == 0.) { *dLdp = 0.; *dLdl = 0.; *dLdw = 0.; } else { NDArray temp = *weightsBroad / sum; *dLdp *= temp; *dLdl *= temp; if(weights->isScalar() || weights->lengthOf() == 1) { *dLdw = 0.; } else { if(weights != weightsBroad) { std::vector axesToReduceAlong = ShapeUtils::evalBroadcastBackwardAxis(weights->getShapeInfo(), weightsBroad->getShapeInfo()); ((E * sum - (E * *weightsBroad).reduceNumber(reduce::Sum)) / (sum*sum)).reduceAlongDimension(reduce::Sum, dLdw, axesToReduceAlong, true, false, false); } else dLdw->assign((E * sum - (E * *weightsBroad).reduceNumber(reduce::Sum)) / (sum*sum)); } } break; } case 3: { // 3 - "weighted_sum_by_nonzero_weights", output is scalar and equal to scalar sum of all elements of E array divided by number of non-zero weights Nd4jLong numOfNonZeroWeights = 0; if(weights->isScalar()) { if(weights->e(0) != 0.) numOfNonZeroWeights = E.lengthOf(); } else numOfNonZeroWeights = weightsBroad->reduceNumber(reduce::CountNonZero).e(0); if (numOfNonZeroWeights == 0) { *dLdp = 0.; *dLdl = 0.; *dLdw = 0.; } else { NDArray temp = *weightsBroad / numOfNonZeroWeights; *dLdp *= temp; *dLdl *= temp; if(weights->isScalar() || weights->lengthOf() == 1) { dLdw->assign(E.reduceNumber(reduce::Sum) / numOfNonZeroWeights); } else { if(weights != weightsBroad) { std::vector axesToReduceAlong = ShapeUtils::evalBroadcastBackwardAxis(weights->getShapeInfo(), weightsBroad->getShapeInfo()); E.reduceAlongDimension(reduce::Sum, dLdw, axesToReduceAlong, true, false, false); *dLdw /= numOfNonZeroWeights; } else dLdw->assign(E / numOfNonZeroWeights); } } break; } } if(weightsBroad != weights) delete weightsBroad; return Status::OK(); } ////////////////////////////////////////////////////////////////////////// DECLARE_TYPES(cosine_distance_loss_grad) { getOpDescriptor()->setAllowedInputTypes(nd4j::DataType::ANY)->setAllowedOutputTypes({ALL_FLOATS}); } ////////////////////////////////////////////////////////////////////////// DECLARE_SHAPE_FN(cosine_distance_loss_grad) { /// labels and predictions must have the same shapes auto predictionsShapeInfo = inputShape->at(0); auto weightsShapeInfo = inputShape->at(1); auto labelsShapeInfo = inputShape->at(2); int dim = INT_ARG(1); if(dim < 0) dim += labelsShapeInfo[0]; std::vector dimensions = {dim}; // labels and predictions must have the same shapes REQUIRE_TRUE(shape::shapeEquals(labelsShapeInfo, predictionsShapeInfo), 0, "COSINE_DISTANCE_LOSS_GRAD OP: labels and predictions arrays must have the same shapes, but got %s and %s correspondingly !", ShapeUtils::shapeAsString(labelsShapeInfo).c_str(), ShapeUtils::shapeAsString(predictionsShapeInfo).c_str()); auto lossShapeInfo = ShapeUtils::evalReduceShapeInfo(shape::order(predictionsShapeInfo), dimensions, predictionsShapeInfo, true, false, block.getWorkspace()); // weights array can be single scalar or has the same rank as loss, and must be broadcastable to loss REQUIRE_TRUE(shape::isScalar(weightsShapeInfo) || shape::rank(weightsShapeInfo) == shape::rank(lossShapeInfo), 0, "COSINE_DISTANCE_LOSS_GRAD OP: weights array should be scalar or have the same rank as loss array, but got %i and %i correspondingly!", shape::rank(weightsShapeInfo), shape::rank(lossShapeInfo)); // check whether broadcast operation is possible for weights array REQUIRE_TRUE(shape::isScalar(weightsShapeInfo) || ShapeUtils::areShapesBroadcastable(weightsShapeInfo, lossShapeInfo), 0, "COSINE_DISTANCE_LOSS_GRAD OP: shapes of weights and loss arrays should be broadcastable, but got weights = %s and loss = %s instead!", ShapeUtils::shapeAsString(weightsShapeInfo).c_str(), ShapeUtils::shapeAsString(lossShapeInfo).c_str()); // input dimension can't be larger than labels/predictions/weights rank REQUIRE_TRUE(dim < labelsShapeInfo[0], 0, "COSINE_DISTANCE_LOSS_GRAD OP: input reduction dimension (got %i) must be < labels rank %i!", dim, labelsShapeInfo[0]); auto outType = DataTypeUtils::pickFloatingType(ArrayOptions::dataType(predictionsShapeInfo)); auto dLdpShapeInfo = ShapeBuilders::copyShapeInfoAndType(predictionsShapeInfo, outType, false, block.getWorkspace()); auto dLdwShapeInfo = ShapeBuilders::copyShapeInfoAndType(weightsShapeInfo, outType, false, block.getWorkspace()); auto dLdlShapeInfo = ShapeBuilders::copyShapeInfoAndType(labelsShapeInfo, outType, false, block.getWorkspace()); return SHAPELIST(CONSTANT(dLdpShapeInfo), CONSTANT(dLdwShapeInfo), CONSTANT(dLdlShapeInfo)); } } } #endif