Skip to content
体验新版
项目
组织
正在加载...
登录
切换导航
打开侧边栏
SummerGao.
Paddle
提交
80c68d38
P
Paddle
项目概览
SummerGao.
/
Paddle
与 Fork 源项目一致
Fork自
PaddlePaddle / Paddle
通知
1
Star
0
Fork
0
代码
文件
提交
分支
Tags
贡献者
分支图
Diff
Issue
0
列表
看板
标记
里程碑
合并请求
0
Wiki
0
Wiki
分析
仓库
DevOps
项目成员
Pages
P
Paddle
项目概览
项目概览
详情
发布
仓库
仓库
文件
提交
分支
标签
贡献者
分支图
比较
Issue
0
Issue
0
列表
看板
标记
里程碑
合并请求
0
合并请求
0
Pages
分析
分析
仓库分析
DevOps
Wiki
0
Wiki
成员
成员
收起侧边栏
关闭侧边栏
动态
分支图
创建新Issue
提交
Issue看板
体验新版 GitCode,发现更多精彩内容 >>
提交
80c68d38
编写于
11月 22, 2016
作者:
L
Luo Tao
浏览文件
操作
浏览文件
下载
电子邮件补丁
差异文件
clang format .cc .h .cpp .c and .hpp file
上级
ad0c144e
变更
377
展开全部
隐藏空白更改
内联
并排
Showing
377 changed file
with
10846 addition
and
7850 deletion
+10846
-7850
paddle/api/Arguments.cpp
paddle/api/Arguments.cpp
+1
-2
paddle/api/ConfigParser.cpp
paddle/api/ConfigParser.cpp
+1
-3
paddle/api/GradientMachine.cpp
paddle/api/GradientMachine.cpp
+16
-9
paddle/api/Internal.h
paddle/api/Internal.h
+4
-4
paddle/api/Matrix.cpp
paddle/api/Matrix.cpp
+34
-17
paddle/api/PaddleAPI.h
paddle/api/PaddleAPI.h
+56
-32
paddle/api/Parameter.cpp
paddle/api/Parameter.cpp
+0
-1
paddle/api/ParameterOptimizer.cpp
paddle/api/ParameterOptimizer.cpp
+21
-15
paddle/api/SequenceGenerator.cpp
paddle/api/SequenceGenerator.cpp
+6
-4
paddle/api/Trainer.cpp
paddle/api/Trainer.cpp
+8
-8
paddle/api/Util.cpp
paddle/api/Util.cpp
+5
-3
paddle/api/Vector.cpp
paddle/api/Vector.cpp
+12
-9
paddle/cuda/include/hl_activation_functions.h
paddle/cuda/include/hl_activation_functions.h
+5
-9
paddle/cuda/include/hl_aggregate.h
paddle/cuda/include/hl_aggregate.h
+0
-1
paddle/cuda/include/hl_avx_functions.h
paddle/cuda/include/hl_avx_functions.h
+9
-10
paddle/cuda/include/hl_base.h
paddle/cuda/include/hl_base.h
+47
-54
paddle/cuda/include/hl_batch_transpose.h
paddle/cuda/include/hl_batch_transpose.h
+2
-6
paddle/cuda/include/hl_cnn.h
paddle/cuda/include/hl_cnn.h
+150
-92
paddle/cuda/include/hl_cuda.h
paddle/cuda/include/hl_cuda.h
+10
-13
paddle/cuda/include/hl_cuda_cublas.h
paddle/cuda/include/hl_cuda_cublas.h
+44
-34
paddle/cuda/include/hl_cuda_cudnn.h
paddle/cuda/include/hl_cuda_cudnn.h
+49
-51
paddle/cuda/include/hl_dso_loader.h
paddle/cuda/include/hl_dso_loader.h
+0
-1
paddle/cuda/include/hl_functions.h
paddle/cuda/include/hl_functions.h
+17
-18
paddle/cuda/include/hl_gpu.h
paddle/cuda/include/hl_gpu.h
+0
-1
paddle/cuda/include/hl_lstm.h
paddle/cuda/include/hl_lstm.h
+0
-1
paddle/cuda/include/hl_matrix.h
paddle/cuda/include/hl_matrix.h
+19
-47
paddle/cuda/include/hl_sequence.h
paddle/cuda/include/hl_sequence.h
+13
-15
paddle/cuda/include/hl_sparse.h
paddle/cuda/include/hl_sparse.h
+41
-37
paddle/cuda/include/hl_table_apply.h
paddle/cuda/include/hl_table_apply.h
+11
-9
paddle/cuda/include/hl_time.h
paddle/cuda/include/hl_time.h
+0
-1
paddle/cuda/include/hl_top_k.h
paddle/cuda/include/hl_top_k.h
+8
-6
paddle/cuda/include/stub/hl_aggregate_stub.h
paddle/cuda/include/stub/hl_aggregate_stub.h
+6
-13
paddle/cuda/include/stub/hl_cnn_stub.h
paddle/cuda/include/stub/hl_cnn_stub.h
+151
-93
paddle/cuda/include/stub/hl_cuda_cublas_stub.h
paddle/cuda/include/stub/hl_cuda_cublas_stub.h
+30
-29
paddle/cuda/include/stub/hl_cuda_cudnn_stub.h
paddle/cuda/include/stub/hl_cuda_cudnn_stub.h
+83
-88
paddle/cuda/include/stub/hl_cuda_stub.h
paddle/cuda/include/stub/hl_cuda_stub.h
+12
-15
paddle/cuda/include/stub/hl_lstm_stub.h
paddle/cuda/include/stub/hl_lstm_stub.h
+0
-1
paddle/cuda/include/stub/hl_matrix_stub.h
paddle/cuda/include/stub/hl_matrix_stub.h
+20
-40
paddle/cuda/include/stub/hl_sequence_stub.h
paddle/cuda/include/stub/hl_sequence_stub.h
+9
-13
paddle/cuda/include/stub/hl_sparse_stub.h
paddle/cuda/include/stub/hl_sparse_stub.h
+41
-39
paddle/cuda/src/avx_mathfun.h
paddle/cuda/src/avx_mathfun.h
+254
-252
paddle/cuda/src/hl_avx_functions.cc
paddle/cuda/src/hl_avx_functions.cc
+40
-44
paddle/cuda/src/hl_cpu_functions.cc
paddle/cuda/src/hl_cpu_functions.cc
+24
-37
paddle/cuda/src/hl_cuda_cublas.cc
paddle/cuda/src/hl_cuda_cublas.cc
+188
-128
paddle/cuda/src/hl_cuda_cudnn.cc
paddle/cuda/src/hl_cuda_cudnn.cc
+626
-620
paddle/cuda/src/hl_cuda_device.cc
paddle/cuda/src/hl_cuda_device.cc
+186
-219
paddle/cuda/src/hl_cudart_wrap.cc
paddle/cuda/src/hl_cudart_wrap.cc
+96
-100
paddle/cuda/src/hl_math.cc
paddle/cuda/src/hl_math.cc
+4
-13
paddle/cuda/src/hl_time.cc
paddle/cuda/src/hl_time.cc
+3
-5
paddle/gserver/activations/ActivationFunction.cpp
paddle/gserver/activations/ActivationFunction.cpp
+46
-24
paddle/gserver/activations/ActivationFunction.h
paddle/gserver/activations/ActivationFunction.h
+0
-1
paddle/gserver/dataproviders/DataProvider.cpp
paddle/gserver/dataproviders/DataProvider.cpp
+13
-8
paddle/gserver/dataproviders/DataProvider.h
paddle/gserver/dataproviders/DataProvider.h
+21
-19
paddle/gserver/dataproviders/DataProviderGroup.h
paddle/gserver/dataproviders/DataProviderGroup.h
+5
-5
paddle/gserver/dataproviders/MultiDataProvider.cpp
paddle/gserver/dataproviders/MultiDataProvider.cpp
+2
-5
paddle/gserver/dataproviders/MultiDataProvider.h
paddle/gserver/dataproviders/MultiDataProvider.h
+0
-1
paddle/gserver/dataproviders/ProtoDataProvider.cpp
paddle/gserver/dataproviders/ProtoDataProvider.cpp
+79
-50
paddle/gserver/dataproviders/ProtoDataProvider.h
paddle/gserver/dataproviders/ProtoDataProvider.h
+6
-4
paddle/gserver/dataproviders/ProtoReader.h
paddle/gserver/dataproviders/ProtoReader.h
+2
-2
paddle/gserver/dataproviders/PyDataProvider.cpp
paddle/gserver/dataproviders/PyDataProvider.cpp
+82
-51
paddle/gserver/dataproviders/PyDataProvider.h
paddle/gserver/dataproviders/PyDataProvider.h
+14
-8
paddle/gserver/dataproviders/PyDataProvider2.cpp
paddle/gserver/dataproviders/PyDataProvider2.cpp
+139
-173
paddle/gserver/evaluators/CTCErrorEvaluator.cpp
paddle/gserver/evaluators/CTCErrorEvaluator.cpp
+11
-7
paddle/gserver/evaluators/ChunkEvaluator.cpp
paddle/gserver/evaluators/ChunkEvaluator.cpp
+2
-1
paddle/gserver/evaluators/Evaluator.cpp
paddle/gserver/evaluators/Evaluator.cpp
+44
-25
paddle/gserver/evaluators/Evaluator.h
paddle/gserver/evaluators/Evaluator.h
+17
-8
paddle/gserver/gradientmachines/GradientMachine.cpp
paddle/gserver/gradientmachines/GradientMachine.cpp
+7
-6
paddle/gserver/gradientmachines/GradientMachine.h
paddle/gserver/gradientmachines/GradientMachine.h
+7
-7
paddle/gserver/gradientmachines/GradientMachineMode.h
paddle/gserver/gradientmachines/GradientMachineMode.h
+22
-21
paddle/gserver/gradientmachines/MultiGradientMachine.cpp
paddle/gserver/gradientmachines/MultiGradientMachine.cpp
+77
-103
paddle/gserver/gradientmachines/MultiGradientMachine.h
paddle/gserver/gradientmachines/MultiGradientMachine.h
+72
-107
paddle/gserver/gradientmachines/MultiNetwork.cpp
paddle/gserver/gradientmachines/MultiNetwork.cpp
+6
-5
paddle/gserver/gradientmachines/MultiNetwork.h
paddle/gserver/gradientmachines/MultiNetwork.h
+7
-6
paddle/gserver/gradientmachines/NeuralNetwork.cpp
paddle/gserver/gradientmachines/NeuralNetwork.cpp
+44
-37
paddle/gserver/gradientmachines/NeuralNetwork.h
paddle/gserver/gradientmachines/NeuralNetwork.h
+11
-11
paddle/gserver/gradientmachines/ParallelNeuralNetwork.cpp
paddle/gserver/gradientmachines/ParallelNeuralNetwork.cpp
+10
-8
paddle/gserver/gradientmachines/ParallelNeuralNetwork.h
paddle/gserver/gradientmachines/ParallelNeuralNetwork.h
+12
-10
paddle/gserver/gradientmachines/RecurrentGradientMachine.cpp
paddle/gserver/gradientmachines/RecurrentGradientMachine.cpp
+91
-54
paddle/gserver/gradientmachines/RecurrentGradientMachine.h
paddle/gserver/gradientmachines/RecurrentGradientMachine.h
+26
-15
paddle/gserver/layers/AddtoLayer.cpp
paddle/gserver/layers/AddtoLayer.cpp
+0
-1
paddle/gserver/layers/AddtoLayer.h
paddle/gserver/layers/AddtoLayer.h
+12
-12
paddle/gserver/layers/AgentLayer.cpp
paddle/gserver/layers/AgentLayer.cpp
+24
-12
paddle/gserver/layers/AgentLayer.h
paddle/gserver/layers/AgentLayer.h
+14
-10
paddle/gserver/layers/AverageLayer.cpp
paddle/gserver/layers/AverageLayer.cpp
+2
-2
paddle/gserver/layers/BatchNormBaseLayer.cpp
paddle/gserver/layers/BatchNormBaseLayer.cpp
+0
-1
paddle/gserver/layers/BatchNormBaseLayer.h
paddle/gserver/layers/BatchNormBaseLayer.h
+6
-7
paddle/gserver/layers/BatchNormalizationLayer.cpp
paddle/gserver/layers/BatchNormalizationLayer.cpp
+43
-34
paddle/gserver/layers/BatchNormalizationLayer.h
paddle/gserver/layers/BatchNormalizationLayer.h
+0
-1
paddle/gserver/layers/BilinearInterpLayer.cpp
paddle/gserver/layers/BilinearInterpLayer.cpp
+21
-9
paddle/gserver/layers/BlockExpandLayer.cpp
paddle/gserver/layers/BlockExpandLayer.cpp
+49
-18
paddle/gserver/layers/BlockExpandLayer.h
paddle/gserver/layers/BlockExpandLayer.h
+0
-1
paddle/gserver/layers/CRFDecodingLayer.cpp
paddle/gserver/layers/CRFDecodingLayer.cpp
+2
-2
paddle/gserver/layers/CRFDecodingLayer.h
paddle/gserver/layers/CRFDecodingLayer.h
+0
-1
paddle/gserver/layers/CRFLayer.cpp
paddle/gserver/layers/CRFLayer.cpp
+8
-8
paddle/gserver/layers/CRFLayer.h
paddle/gserver/layers/CRFLayer.h
+1
-2
paddle/gserver/layers/CTCLayer.cpp
paddle/gserver/layers/CTCLayer.cpp
+11
-14
paddle/gserver/layers/CTCLayer.h
paddle/gserver/layers/CTCLayer.h
+2
-2
paddle/gserver/layers/ConcatenateLayer.cpp
paddle/gserver/layers/ConcatenateLayer.cpp
+3
-4
paddle/gserver/layers/ContextProjection.cpp
paddle/gserver/layers/ContextProjection.cpp
+30
-16
paddle/gserver/layers/ContextProjection.h
paddle/gserver/layers/ContextProjection.h
+2
-2
paddle/gserver/layers/ConvBaseLayer.cpp
paddle/gserver/layers/ConvBaseLayer.cpp
+21
-28
paddle/gserver/layers/ConvBaseLayer.h
paddle/gserver/layers/ConvBaseLayer.h
+0
-1
paddle/gserver/layers/ConvOperator.cpp
paddle/gserver/layers/ConvOperator.cpp
+76
-25
paddle/gserver/layers/ConvProjection.cpp
paddle/gserver/layers/ConvProjection.cpp
+86
-42
paddle/gserver/layers/ConvProjection.h
paddle/gserver/layers/ConvProjection.h
+10
-3
paddle/gserver/layers/ConvShiftLayer.cpp
paddle/gserver/layers/ConvShiftLayer.cpp
+0
-1
paddle/gserver/layers/ConvexCombinationLayer.cpp
paddle/gserver/layers/ConvexCombinationLayer.cpp
+15
-7
paddle/gserver/layers/CosSimLayer.cpp
paddle/gserver/layers/CosSimLayer.cpp
+6
-4
paddle/gserver/layers/CosSimLayer.h
paddle/gserver/layers/CosSimLayer.h
+1
-3
paddle/gserver/layers/CosSimVecMatLayer.cpp
paddle/gserver/layers/CosSimVecMatLayer.cpp
+37
-16
paddle/gserver/layers/CostLayer.cpp
paddle/gserver/layers/CostLayer.cpp
+70
-49
paddle/gserver/layers/CostLayer.h
paddle/gserver/layers/CostLayer.h
+8
-3
paddle/gserver/layers/CudnnBatchNormLayer.cpp
paddle/gserver/layers/CudnnBatchNormLayer.cpp
+35
-12
paddle/gserver/layers/CudnnBatchNormLayer.h
paddle/gserver/layers/CudnnBatchNormLayer.h
+2
-2
paddle/gserver/layers/CudnnConvLayer.cpp
paddle/gserver/layers/CudnnConvLayer.cpp
+15
-9
paddle/gserver/layers/CudnnConvLayer.h
paddle/gserver/layers/CudnnConvLayer.h
+0
-1
paddle/gserver/layers/CudnnPoolLayer.cpp
paddle/gserver/layers/CudnnPoolLayer.cpp
+18
-5
paddle/gserver/layers/CudnnPoolLayer.h
paddle/gserver/layers/CudnnPoolLayer.h
+6
-7
paddle/gserver/layers/DataLayer.cpp
paddle/gserver/layers/DataLayer.cpp
+7
-6
paddle/gserver/layers/DataLayer.h
paddle/gserver/layers/DataLayer.h
+4
-7
paddle/gserver/layers/DataNormLayer.cpp
paddle/gserver/layers/DataNormLayer.cpp
+22
-11
paddle/gserver/layers/DataNormLayer.h
paddle/gserver/layers/DataNormLayer.h
+0
-1
paddle/gserver/layers/DotMulOperator.cpp
paddle/gserver/layers/DotMulOperator.cpp
+2
-3
paddle/gserver/layers/DotMulProjection.cpp
paddle/gserver/layers/DotMulProjection.cpp
+4
-3
paddle/gserver/layers/EosIdCheckLayer.cpp
paddle/gserver/layers/EosIdCheckLayer.cpp
+1
-2
paddle/gserver/layers/ExpandConvBaseLayer.cpp
paddle/gserver/layers/ExpandConvBaseLayer.cpp
+67
-38
paddle/gserver/layers/ExpandConvBaseLayer.h
paddle/gserver/layers/ExpandConvBaseLayer.h
+1
-2
paddle/gserver/layers/ExpandConvLayer.cpp
paddle/gserver/layers/ExpandConvLayer.cpp
+0
-2
paddle/gserver/layers/ExpandConvLayer.h
paddle/gserver/layers/ExpandConvLayer.h
+2
-3
paddle/gserver/layers/ExpandConvTransLayer.cpp
paddle/gserver/layers/ExpandConvTransLayer.cpp
+1
-3
paddle/gserver/layers/ExpandConvTransLayer.h
paddle/gserver/layers/ExpandConvTransLayer.h
+2
-3
paddle/gserver/layers/FeatureMapExpandLayer.cpp
paddle/gserver/layers/FeatureMapExpandLayer.cpp
+12
-7
paddle/gserver/layers/FullMatrixProjection.cpp
paddle/gserver/layers/FullMatrixProjection.cpp
+0
-1
paddle/gserver/layers/FullMatrixProjection.h
paddle/gserver/layers/FullMatrixProjection.h
+2
-1
paddle/gserver/layers/FullyConnectedLayer.cpp
paddle/gserver/layers/FullyConnectedLayer.cpp
+0
-1
paddle/gserver/layers/FullyConnectedLayer.h
paddle/gserver/layers/FullyConnectedLayer.h
+3
-5
paddle/gserver/layers/GatedRecurrentLayer.cpp
paddle/gserver/layers/GatedRecurrentLayer.cpp
+53
-49
paddle/gserver/layers/GatedRecurrentLayer.h
paddle/gserver/layers/GatedRecurrentLayer.h
+13
-7
paddle/gserver/layers/GetOutputLayer.cpp
paddle/gserver/layers/GetOutputLayer.cpp
+0
-1
paddle/gserver/layers/GruCompute.cpp
paddle/gserver/layers/GruCompute.cpp
+14
-17
paddle/gserver/layers/GruCompute.h
paddle/gserver/layers/GruCompute.h
+3
-2
paddle/gserver/layers/GruStepLayer.cpp
paddle/gserver/layers/GruStepLayer.cpp
+19
-12
paddle/gserver/layers/HierarchicalSigmoidLayer.cpp
paddle/gserver/layers/HierarchicalSigmoidLayer.cpp
+19
-12
paddle/gserver/layers/HierarchicalSigmoidLayer.h
paddle/gserver/layers/HierarchicalSigmoidLayer.h
+7
-8
paddle/gserver/layers/IdentityProjection.cpp
paddle/gserver/layers/IdentityProjection.cpp
+4
-3
paddle/gserver/layers/InterpolationLayer.cpp
paddle/gserver/layers/InterpolationLayer.cpp
+2
-3
paddle/gserver/layers/Layer.cpp
paddle/gserver/layers/Layer.cpp
+31
-23
paddle/gserver/layers/Layer.h
paddle/gserver/layers/Layer.h
+21
-18
paddle/gserver/layers/LinearChainCRF.cpp
paddle/gserver/layers/LinearChainCRF.cpp
+0
-1
paddle/gserver/layers/LinearChainCRF.h
paddle/gserver/layers/LinearChainCRF.h
+2
-2
paddle/gserver/layers/LinearChainCTC.cpp
paddle/gserver/layers/LinearChainCTC.cpp
+9
-6
paddle/gserver/layers/LinearChainCTC.h
paddle/gserver/layers/LinearChainCTC.h
+6
-3
paddle/gserver/layers/LstmCompute.cpp
paddle/gserver/layers/LstmCompute.cpp
+21
-11
paddle/gserver/layers/LstmCompute.h
paddle/gserver/layers/LstmCompute.h
+5
-2
paddle/gserver/layers/LstmLayer.cpp
paddle/gserver/layers/LstmLayer.cpp
+183
-78
paddle/gserver/layers/LstmLayer.h
paddle/gserver/layers/LstmLayer.h
+19
-7
paddle/gserver/layers/LstmStepLayer.cpp
paddle/gserver/layers/LstmStepLayer.cpp
+46
-28
paddle/gserver/layers/MDLstmLayer.cpp
paddle/gserver/layers/MDLstmLayer.cpp
+174
-81
paddle/gserver/layers/MaxIdLayer.cpp
paddle/gserver/layers/MaxIdLayer.cpp
+4
-1
paddle/gserver/layers/MaxLayer.cpp
paddle/gserver/layers/MaxLayer.cpp
+2
-2
paddle/gserver/layers/MaxLayer.h
paddle/gserver/layers/MaxLayer.h
+0
-1
paddle/gserver/layers/MixedLayer.cpp
paddle/gserver/layers/MixedLayer.cpp
+3
-5
paddle/gserver/layers/MixedLayer.h
paddle/gserver/layers/MixedLayer.h
+3
-4
paddle/gserver/layers/MultinomialSampler.cpp
paddle/gserver/layers/MultinomialSampler.cpp
+0
-1
paddle/gserver/layers/MultinomialSampler.h
paddle/gserver/layers/MultinomialSampler.h
+0
-1
paddle/gserver/layers/MultiplexLayer.cpp
paddle/gserver/layers/MultiplexLayer.cpp
+0
-1
paddle/gserver/layers/NCELayer.cpp
paddle/gserver/layers/NCELayer.cpp
+20
-8
paddle/gserver/layers/NormLayer.cpp
paddle/gserver/layers/NormLayer.cpp
+0
-1
paddle/gserver/layers/NormLayer.h
paddle/gserver/layers/NormLayer.h
+2
-3
paddle/gserver/layers/NormProjectionLayer.cpp
paddle/gserver/layers/NormProjectionLayer.cpp
+11
-5
paddle/gserver/layers/NormProjectionLayer.h
paddle/gserver/layers/NormProjectionLayer.h
+0
-1
paddle/gserver/layers/Operator.cpp
paddle/gserver/layers/Operator.cpp
+0
-1
paddle/gserver/layers/Operator.h
paddle/gserver/layers/Operator.h
+4
-3
paddle/gserver/layers/OuterProdLayer.cpp
paddle/gserver/layers/OuterProdLayer.cpp
+8
-6
paddle/gserver/layers/ParameterReluLayer.cpp
paddle/gserver/layers/ParameterReluLayer.cpp
+2
-3
paddle/gserver/layers/ParameterReluLayer.h
paddle/gserver/layers/ParameterReluLayer.h
+0
-1
paddle/gserver/layers/PoolLayer.cpp
paddle/gserver/layers/PoolLayer.cpp
+0
-1
paddle/gserver/layers/PoolLayer.h
paddle/gserver/layers/PoolLayer.h
+0
-1
paddle/gserver/layers/PoolProjection.cpp
paddle/gserver/layers/PoolProjection.cpp
+62
-14
paddle/gserver/layers/PoolProjection.h
paddle/gserver/layers/PoolProjection.h
+8
-4
paddle/gserver/layers/PoolProjectionLayer.cpp
paddle/gserver/layers/PoolProjectionLayer.cpp
+8
-3
paddle/gserver/layers/PowerLayer.cpp
paddle/gserver/layers/PowerLayer.cpp
+1
-2
paddle/gserver/layers/PrintLayer.cpp
paddle/gserver/layers/PrintLayer.cpp
+1
-2
paddle/gserver/layers/Projection.cpp
paddle/gserver/layers/Projection.cpp
+2
-2
paddle/gserver/layers/Projection.h
paddle/gserver/layers/Projection.h
+4
-2
paddle/gserver/layers/RecurrentLayer.cpp
paddle/gserver/layers/RecurrentLayer.cpp
+40
-23
paddle/gserver/layers/RecurrentLayerGroup.cpp
paddle/gserver/layers/RecurrentLayerGroup.cpp
+7
-5
paddle/gserver/layers/ResizeLayer.cpp
paddle/gserver/layers/ResizeLayer.cpp
+5
-4
paddle/gserver/layers/ScalingLayer.cpp
paddle/gserver/layers/ScalingLayer.cpp
+1
-2
paddle/gserver/layers/ScalingProjection.cpp
paddle/gserver/layers/ScalingProjection.cpp
+8
-4
paddle/gserver/layers/SelectiveFullyConnectedLayer.cpp
paddle/gserver/layers/SelectiveFullyConnectedLayer.cpp
+78
-35
paddle/gserver/layers/SelectiveFullyConnectedLayer.h
paddle/gserver/layers/SelectiveFullyConnectedLayer.h
+1
-4
paddle/gserver/layers/SequenceConcatLayer.cpp
paddle/gserver/layers/SequenceConcatLayer.cpp
+6
-11
paddle/gserver/layers/SequenceLastInstanceLayer.cpp
paddle/gserver/layers/SequenceLastInstanceLayer.cpp
+0
-1
paddle/gserver/layers/SequencePoolLayer.cpp
paddle/gserver/layers/SequencePoolLayer.cpp
+1
-1
paddle/gserver/layers/SequenceReshapeLayer.cpp
paddle/gserver/layers/SequenceReshapeLayer.cpp
+7
-8
paddle/gserver/layers/SequenceToBatch.cpp
paddle/gserver/layers/SequenceToBatch.cpp
+36
-20
paddle/gserver/layers/SequenceToBatch.h
paddle/gserver/layers/SequenceToBatch.h
+11
-5
paddle/gserver/layers/SlopeInterceptLayer.cpp
paddle/gserver/layers/SlopeInterceptLayer.cpp
+4
-3
paddle/gserver/layers/SpatialPyramidPoolLayer.cpp
paddle/gserver/layers/SpatialPyramidPoolLayer.cpp
+2
-1
paddle/gserver/layers/SpatialPyramidPoolLayer.h
paddle/gserver/layers/SpatialPyramidPoolLayer.h
+6
-3
paddle/gserver/layers/SubSequenceLayer.cpp
paddle/gserver/layers/SubSequenceLayer.cpp
+6
-11
paddle/gserver/layers/SumToOneNormLayer.cpp
paddle/gserver/layers/SumToOneNormLayer.cpp
+1
-2
paddle/gserver/layers/TableProjection.cpp
paddle/gserver/layers/TableProjection.cpp
+2
-2
paddle/gserver/layers/TableProjection.h
paddle/gserver/layers/TableProjection.h
+2
-2
paddle/gserver/layers/TensorLayer.cpp
paddle/gserver/layers/TensorLayer.cpp
+6
-3
paddle/gserver/layers/TensorLayer.h
paddle/gserver/layers/TensorLayer.h
+0
-1
paddle/gserver/layers/TransLayer.cpp
paddle/gserver/layers/TransLayer.cpp
+0
-1
paddle/gserver/layers/TransLayer.h
paddle/gserver/layers/TransLayer.h
+0
-1
paddle/gserver/layers/TransposedFullMatrixProjection.cpp
paddle/gserver/layers/TransposedFullMatrixProjection.cpp
+2
-2
paddle/gserver/layers/ValidationLayer.cpp
paddle/gserver/layers/ValidationLayer.cpp
+5
-3
paddle/gserver/tests/LayerGradUtil.cpp
paddle/gserver/tests/LayerGradUtil.cpp
+135
-57
paddle/gserver/tests/LayerGradUtil.h
paddle/gserver/tests/LayerGradUtil.h
+73
-30
paddle/gserver/tests/TestUtil.cpp
paddle/gserver/tests/TestUtil.cpp
+13
-13
paddle/gserver/tests/TestUtil.h
paddle/gserver/tests/TestUtil.h
+13
-14
paddle/gserver/tests/test_ActivationGrad.cpp
paddle/gserver/tests/test_ActivationGrad.cpp
+2
-2
paddle/gserver/tests/test_ConvTrans.cpp
paddle/gserver/tests/test_ConvTrans.cpp
+193
-193
paddle/gserver/tests/test_Evaluator.cpp
paddle/gserver/tests/test_Evaluator.cpp
+15
-11
paddle/gserver/tests/test_LayerGrad.cpp
paddle/gserver/tests/test_LayerGrad.cpp
+176
-76
paddle/gserver/tests/test_LinearChainCRF.cpp
paddle/gserver/tests/test_LinearChainCRF.cpp
+0
-1
paddle/gserver/tests/test_MultinomialSampler.cpp
paddle/gserver/tests/test_MultinomialSampler.cpp
+1
-3
paddle/gserver/tests/test_NetworkCompare.cpp
paddle/gserver/tests/test_NetworkCompare.cpp
+20
-11
paddle/gserver/tests/test_ProtoDataProvider.cpp
paddle/gserver/tests/test_ProtoDataProvider.cpp
+51
-34
paddle/gserver/tests/test_PyDataProvider.cpp
paddle/gserver/tests/test_PyDataProvider.cpp
+4
-4
paddle/gserver/tests/test_PyDataProvider2.cpp
paddle/gserver/tests/test_PyDataProvider2.cpp
+48
-48
paddle/gserver/tests/test_RecurrentGradientMachine.cpp
paddle/gserver/tests/test_RecurrentGradientMachine.cpp
+20
-12
paddle/gserver/tests/test_RecurrentLayer.cpp
paddle/gserver/tests/test_RecurrentLayer.cpp
+55
-32
paddle/gserver/tests/test_SelectiveFCLayer.cpp
paddle/gserver/tests/test_SelectiveFCLayer.cpp
+92
-66
paddle/math/Allocator.h
paddle/math/Allocator.h
+11
-16
paddle/math/BaseMatrix.h
paddle/math/BaseMatrix.h
+99
-46
paddle/math/CpuSparseMatrix.cpp
paddle/math/CpuSparseMatrix.cpp
+107
-42
paddle/math/CpuSparseMatrix.h
paddle/math/CpuSparseMatrix.h
+30
-18
paddle/math/ExecViaCpu.h
paddle/math/ExecViaCpu.h
+10
-6
paddle/math/MathFunctions.cpp
paddle/math/MathFunctions.cpp
+128
-69
paddle/math/MathFunctions.h
paddle/math/MathFunctions.h
+36
-24
paddle/math/MathUtils.cpp
paddle/math/MathUtils.cpp
+11
-10
paddle/math/MathUtils.h
paddle/math/MathUtils.h
+6
-6
paddle/math/Matrix.cpp
paddle/math/Matrix.cpp
+797
-346
paddle/math/Matrix.h
paddle/math/Matrix.h
+554
-245
paddle/math/MatrixBitCode.cpp
paddle/math/MatrixBitCode.cpp
+47
-29
paddle/math/MemoryHandle.cpp
paddle/math/MemoryHandle.cpp
+3
-8
paddle/math/MemoryHandle.h
paddle/math/MemoryHandle.h
+3
-4
paddle/math/PoolAllocator.cpp
paddle/math/PoolAllocator.cpp
+3
-5
paddle/math/PoolAllocator.h
paddle/math/PoolAllocator.h
+0
-1
paddle/math/SIMDFunctions.cpp
paddle/math/SIMDFunctions.cpp
+10
-8
paddle/math/SIMDFunctions.h
paddle/math/SIMDFunctions.h
+4
-6
paddle/math/SparseMatrix.cpp
paddle/math/SparseMatrix.cpp
+178
-85
paddle/math/SparseMatrix.h
paddle/math/SparseMatrix.h
+46
-23
paddle/math/SparseRowMatrix.cpp
paddle/math/SparseRowMatrix.cpp
+41
-32
paddle/math/SparseRowMatrix.h
paddle/math/SparseRowMatrix.h
+31
-16
paddle/math/Storage.cpp
paddle/math/Storage.cpp
+11
-12
paddle/math/Vector.cpp
paddle/math/Vector.cpp
+75
-57
paddle/math/Vector.h
paddle/math/Vector.h
+19
-26
paddle/math/tests/test_Allocator.cpp
paddle/math/tests/test_Allocator.cpp
+4
-4
paddle/math/tests/test_ExecViaCpu.cpp
paddle/math/tests/test_ExecViaCpu.cpp
+16
-6
paddle/math/tests/test_FPException.cpp
paddle/math/tests/test_FPException.cpp
+5
-6
paddle/math/tests/test_SIMDFunctions.cpp
paddle/math/tests/test_SIMDFunctions.cpp
+4
-6
paddle/math/tests/test_batchTranspose.cpp
paddle/math/tests/test_batchTranspose.cpp
+2
-3
paddle/math/tests/test_matrix.cpp
paddle/math/tests/test_matrix.cpp
+95
-41
paddle/math/tests/test_matrixCompare.cpp
paddle/math/tests/test_matrixCompare.cpp
+302
-164
paddle/math/tests/test_matrixUtil.h
paddle/math/tests/test_matrixUtil.h
+3
-3
paddle/math/tests/test_perturbation.cpp
paddle/math/tests/test_perturbation.cpp
+120
-54
paddle/math/tests/test_sparseMatrixCompare.cpp
paddle/math/tests/test_sparseMatrixCompare.cpp
+1
-1
paddle/parameter/Argument.cpp
paddle/parameter/Argument.cpp
+136
-83
paddle/parameter/Argument.h
paddle/parameter/Argument.h
+25
-14
paddle/parameter/AverageOptimizer.cpp
paddle/parameter/AverageOptimizer.cpp
+26
-14
paddle/parameter/AverageOptimizer.h
paddle/parameter/AverageOptimizer.h
+10
-6
paddle/parameter/FirstOrderOptimizer.cpp
paddle/parameter/FirstOrderOptimizer.cpp
+48
-33
paddle/parameter/FirstOrderOptimizer.h
paddle/parameter/FirstOrderOptimizer.h
+29
-19
paddle/parameter/LearningRateScheduler.cpp
paddle/parameter/LearningRateScheduler.cpp
+0
-1
paddle/parameter/LearningRateScheduler.h
paddle/parameter/LearningRateScheduler.h
+4
-4
paddle/parameter/OptimizerFunctions.cpp
paddle/parameter/OptimizerFunctions.cpp
+9
-7
paddle/parameter/OptimizerFunctions.h
paddle/parameter/OptimizerFunctions.h
+2
-2
paddle/parameter/OptimizerWithRegularizer.cpp
paddle/parameter/OptimizerWithRegularizer.cpp
+35
-24
paddle/parameter/OptimizerWithRegularizer.h
paddle/parameter/OptimizerWithRegularizer.h
+12
-7
paddle/parameter/ParallelParameter.cpp
paddle/parameter/ParallelParameter.cpp
+2
-2
paddle/parameter/ParallelParameter.h
paddle/parameter/ParallelParameter.h
+10
-10
paddle/parameter/Parameter.cpp
paddle/parameter/Parameter.cpp
+88
-51
paddle/parameter/Parameter.h
paddle/parameter/Parameter.h
+10
-8
paddle/parameter/ParameterOptimizer.cpp
paddle/parameter/ParameterOptimizer.cpp
+0
-1
paddle/parameter/ParameterOptimizer.h
paddle/parameter/ParameterOptimizer.h
+70
-69
paddle/parameter/ParameterUpdateFunctions.cpp
paddle/parameter/ParameterUpdateFunctions.cpp
+43
-13
paddle/parameter/ParameterUpdateFunctions.h
paddle/parameter/ParameterUpdateFunctions.h
+21
-9
paddle/parameter/ParameterUpdaterBase.cpp
paddle/parameter/ParameterUpdaterBase.cpp
+0
-1
paddle/parameter/ParameterUpdaterBase.h
paddle/parameter/ParameterUpdaterBase.h
+0
-1
paddle/parameter/ParameterUpdaterHook.cpp
paddle/parameter/ParameterUpdaterHook.cpp
+2
-2
paddle/parameter/ParameterUpdaterHook.h
paddle/parameter/ParameterUpdaterHook.h
+0
-1
paddle/parameter/Regularizer.cpp
paddle/parameter/Regularizer.cpp
+3
-3
paddle/parameter/Regularizer.h
paddle/parameter/Regularizer.h
+32
-14
paddle/parameter/Weight.cpp
paddle/parameter/Weight.cpp
+10
-4
paddle/parameter/tests/test_common.cpp
paddle/parameter/tests/test_common.cpp
+43
-29
paddle/pserver/BaseClient.cpp
paddle/pserver/BaseClient.cpp
+0
-1
paddle/pserver/BaseClient.h
paddle/pserver/BaseClient.h
+32
-11
paddle/pserver/LightNetwork.cpp
paddle/pserver/LightNetwork.cpp
+25
-18
paddle/pserver/LightNetwork.h
paddle/pserver/LightNetwork.h
+5
-6
paddle/pserver/ParameterClient2.cpp
paddle/pserver/ParameterClient2.cpp
+74
-41
paddle/pserver/ParameterClient2.h
paddle/pserver/ParameterClient2.h
+64
-33
paddle/pserver/ParameterServer2.cpp
paddle/pserver/ParameterServer2.cpp
+97
-60
paddle/pserver/ParameterServer2.h
paddle/pserver/ParameterServer2.h
+29
-21
paddle/pserver/ProtoServer.cpp
paddle/pserver/ProtoServer.cpp
+2
-3
paddle/pserver/ProtoServer.h
paddle/pserver/ProtoServer.h
+82
-67
paddle/pserver/RDMANetwork.h
paddle/pserver/RDMANetwork.h
+1
-3
paddle/pserver/SocketChannel.cpp
paddle/pserver/SocketChannel.cpp
+32
-15
paddle/pserver/SocketChannel.h
paddle/pserver/SocketChannel.h
+0
-1
paddle/pserver/SparseParameterDistribution.cpp
paddle/pserver/SparseParameterDistribution.cpp
+10
-6
paddle/pserver/test/SocketTest.cpp
paddle/pserver/test/SocketTest.cpp
+2
-2
paddle/pserver/test/test_ParameterServer2.cpp
paddle/pserver/test/test_ParameterServer2.cpp
+8
-6
paddle/pserver/test/test_ProtoServer.cpp
paddle/pserver/test/test_ProtoServer.cpp
+5
-3
paddle/trainer/ParamUtil.cpp
paddle/trainer/ParamUtil.cpp
+11
-10
paddle/trainer/ParamUtil.h
paddle/trainer/ParamUtil.h
+12
-13
paddle/trainer/ParameterUpdater.cpp
paddle/trainer/ParameterUpdater.cpp
+2
-2
paddle/trainer/ParameterUpdater.h
paddle/trainer/ParameterUpdater.h
+6
-7
paddle/trainer/RemoteParameterUpdater.cpp
paddle/trainer/RemoteParameterUpdater.cpp
+27
-17
paddle/trainer/RemoteParameterUpdater.h
paddle/trainer/RemoteParameterUpdater.h
+15
-11
paddle/trainer/Tester.cpp
paddle/trainer/Tester.cpp
+45
-52
paddle/trainer/Tester.h
paddle/trainer/Tester.h
+4
-7
paddle/trainer/TesterConfig.h
paddle/trainer/TesterConfig.h
+0
-1
paddle/trainer/ThreadParameterUpdater.cpp
paddle/trainer/ThreadParameterUpdater.cpp
+25
-24
paddle/trainer/ThreadParameterUpdater.h
paddle/trainer/ThreadParameterUpdater.h
+4
-6
paddle/trainer/Trainer.cpp
paddle/trainer/Trainer.cpp
+81
-74
paddle/trainer/Trainer.h
paddle/trainer/Trainer.h
+9
-10
paddle/trainer/TrainerConfigHelper.cpp
paddle/trainer/TrainerConfigHelper.cpp
+14
-23
paddle/trainer/TrainerConfigHelper.h
paddle/trainer/TrainerConfigHelper.h
+4
-15
paddle/trainer/TrainerInternal.cpp
paddle/trainer/TrainerInternal.cpp
+54
-56
paddle/trainer/TrainerInternal.h
paddle/trainer/TrainerInternal.h
+9
-17
paddle/trainer/TrainerInternalConfig.cpp
paddle/trainer/TrainerInternalConfig.cpp
+2
-1
paddle/trainer/TrainerInternalConfig.h
paddle/trainer/TrainerInternalConfig.h
+4
-9
paddle/trainer/TrainerMain.cpp
paddle/trainer/TrainerMain.cpp
+1
-2
paddle/trainer/tests/picojson.h
paddle/trainer/tests/picojson.h
+14
-6
paddle/trainer/tests/test_Compare.cpp
paddle/trainer/tests/test_Compare.cpp
+2
-2
paddle/trainer/tests/test_CompareSparse.cpp
paddle/trainer/tests/test_CompareSparse.cpp
+21
-17
paddle/trainer/tests/test_CompareTwoNets.cpp
paddle/trainer/tests/test_CompareTwoNets.cpp
+28
-12
paddle/trainer/tests/test_CompareTwoOpts.cpp
paddle/trainer/tests/test_CompareTwoOpts.cpp
+25
-9
paddle/trainer/tests/test_Prediction.cpp
paddle/trainer/tests/test_Prediction.cpp
+8
-5
paddle/trainer/tests/test_PyDataProviderWrapper.cpp
paddle/trainer/tests/test_PyDataProviderWrapper.cpp
+0
-1
paddle/trainer/tests/test_Trainer.cpp
paddle/trainer/tests/test_Trainer.cpp
+4
-2
paddle/trainer/tests/test_TrainerOnePass.cpp
paddle/trainer/tests/test_TrainerOnePass.cpp
+19
-16
paddle/trainer/tests/test_recurrent_machine_generation.cpp
paddle/trainer/tests/test_recurrent_machine_generation.cpp
+14
-6
paddle/utils/BarrierStat.cpp
paddle/utils/BarrierStat.cpp
+11
-7
paddle/utils/BarrierStat.h
paddle/utils/BarrierStat.h
+54
-55
paddle/utils/ClassRegistrar.h
paddle/utils/ClassRegistrar.h
+5
-6
paddle/utils/CommandLineParser.cpp
paddle/utils/CommandLineParser.cpp
+19
-25
paddle/utils/CommandLineParser.h
paddle/utils/CommandLineParser.h
+4
-3
paddle/utils/CustomStackTrace.cpp
paddle/utils/CustomStackTrace.cpp
+19
-17
paddle/utils/CustomStackTrace.h
paddle/utils/CustomStackTrace.h
+12
-15
paddle/utils/DisableCopy.h
paddle/utils/DisableCopy.h
+3
-4
paddle/utils/Excepts.cpp
paddle/utils/Excepts.cpp
+6
-6
paddle/utils/Flags.cpp
paddle/utils/Flags.cpp
+20
-11
paddle/utils/Flags.h
paddle/utils/Flags.h
+0
-1
paddle/utils/GlobalConstants.cpp
paddle/utils/GlobalConstants.cpp
+0
-1
paddle/utils/GlobalConstants.h
paddle/utils/GlobalConstants.h
+4
-5
paddle/utils/Locks.h
paddle/utils/Locks.h
+24
-25
paddle/utils/Logging.cpp
paddle/utils/Logging.cpp
+15
-11
paddle/utils/Logging.h
paddle/utils/Logging.h
+4
-4
paddle/utils/PythonUtil.cpp
paddle/utils/PythonUtil.cpp
+17
-22
paddle/utils/PythonUtil.h
paddle/utils/PythonUtil.h
+19
-32
paddle/utils/Queue.h
paddle/utils/Queue.h
+7
-10
paddle/utils/Stat.h
paddle/utils/Stat.h
+16
-13
paddle/utils/StringUtil.h
paddle/utils/StringUtil.h
+0
-3
paddle/utils/Thread.h
paddle/utils/Thread.h
+31
-23
paddle/utils/ThreadLocal.cpp
paddle/utils/ThreadLocal.cpp
+2
-1
paddle/utils/ThreadLocal.h
paddle/utils/ThreadLocal.h
+1
-4
paddle/utils/TypeDefs.h
paddle/utils/TypeDefs.h
+0
-1
paddle/utils/Util.cpp
paddle/utils/Util.cpp
+37
-33
paddle/utils/Util.h
paddle/utils/Util.h
+23
-31
paddle/utils/Version.cpp
paddle/utils/Version.cpp
+14
-11
paddle/utils/Version.h
paddle/utils/Version.h
+3
-13
paddle/utils/arch/linux/Locks.cpp
paddle/utils/arch/linux/Locks.cpp
+9
-23
paddle/utils/arch/osx/Locks.cpp
paddle/utils/arch/osx/Locks.cpp
+10
-18
paddle/utils/tests/test_CommandLineParser.cpp
paddle/utils/tests/test_CommandLineParser.cpp
+10
-7
paddle/utils/tests/test_CustomStackTrace.cpp
paddle/utils/tests/test_CustomStackTrace.cpp
+25
-18
paddle/utils/tests/test_CustomStackTracePrint.cpp
paddle/utils/tests/test_CustomStackTracePrint.cpp
+1
-1
paddle/utils/tests/test_Logging.cpp
paddle/utils/tests/test_Logging.cpp
+2
-4
paddle/utils/tests/test_SpinLock.cpp
paddle/utils/tests/test_SpinLock.cpp
+13
-11
paddle/utils/tests/test_StringUtils.cpp
paddle/utils/tests/test_StringUtils.cpp
+0
-1
paddle/utils/tests/test_Thread.cpp
paddle/utils/tests/test_Thread.cpp
+14
-17
paddle/utils/tests/test_ThreadBarrier.cpp
paddle/utils/tests/test_ThreadBarrier.cpp
+31
-29
未找到文件。
paddle/api/Arguments.cpp
浏览文件 @
80c68d38
...
...
@@ -12,7 +12,6 @@ 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. */
#include "PaddleAPI.h"
#include "PaddleAPIPrivate.h"
...
...
@@ -112,7 +111,7 @@ void Arguments::setSlotSequenceStartPositions(size_t idx,
}
void
Arguments
::
setSlotSubSequenceStartPositions
(
size_t
idx
,
IVector
*
vec
)
throw
(
RangeError
)
{
size_t
idx
,
IVector
*
vec
)
throw
(
RangeError
)
{
auto
&
a
=
m
->
getArg
(
idx
);
auto
&
v
=
m
->
cast
<
paddle
::
IVector
>
(
vec
->
getSharedPtr
());
a
.
subSequenceStartPositions
=
std
::
make_shared
<
paddle
::
ICpuGpuVector
>
(
v
);
...
...
paddle/api/ConfigParser.cpp
浏览文件 @
80c68d38
...
...
@@ -12,7 +12,6 @@ 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. */
#include "PaddleAPI.h"
#include "PaddleAPIPrivate.h"
#include "paddle/trainer/Trainer.h"
...
...
@@ -44,8 +43,7 @@ TrainerConfig* TrainerConfig::createFromTrainerConfigFile(
return
retv
;
}
TrainerConfig
*
TrainerConfig
::
createFromProtoString
(
const
std
::
string
&
str
)
{
TrainerConfig
*
TrainerConfig
::
createFromProtoString
(
const
std
::
string
&
str
)
{
auto
retv
=
new
TrainerConfig
();
paddle
::
TrainerConfig
trainerConfigProto
;
auto
conf
=
std
::
make_shared
<
paddle
::
TrainerConfigHelper
>
(
trainerConfigProto
);
...
...
paddle/api/GradientMachine.cpp
浏览文件 @
80c68d38
...
...
@@ -12,7 +12,6 @@ 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. */
#include "PaddleAPI.h"
#include "PaddleAPIPrivate.h"
...
...
@@ -27,7 +26,8 @@ GradientMachine::GradientMachine() : m(new GradientMachinePrivate()) {}
GradientMachine
::~
GradientMachine
()
{
delete
m
;
}
GradientMachine
*
GradientMachine
::
createFromPaddleModelPtr
(
const
void
*
confPtr
,
GradientMatchineCreateMode
mode
,
const
void
*
confPtr
,
GradientMatchineCreateMode
mode
,
const
std
::
vector
<
int
>&
types
)
{
auto
&
conf
=
*
(
const
paddle
::
ModelConfig
*
)(
confPtr
);
std
::
vector
<
ParameterType
>
realTypes
;
...
...
@@ -44,7 +44,8 @@ GradientMachine* GradientMachine::createFromPaddleModelPtr(
}
GradientMachine
*
GradientMachine
::
createByConfigProtoStr
(
const
std
::
string
&
protoStr
,
GradientMatchineCreateMode
mode
,
const
std
::
string
&
protoStr
,
GradientMatchineCreateMode
mode
,
const
std
::
vector
<
int
>&
types
)
{
paddle
::
ModelConfig
conf
;
conf
.
ParseFromString
(
protoStr
);
...
...
@@ -56,13 +57,15 @@ GradientMachine* GradientMachine::createByConfigProtoStr(
}
GradientMachine
*
GradientMachine
::
createByModelConfig
(
ModelConfig
*
conf
,
GradientMatchineCreateMode
mode
,
ModelConfig
*
conf
,
GradientMatchineCreateMode
mode
,
const
std
::
vector
<
int
>&
types
)
{
auto
confPtr
=
&
conf
->
m
->
conf
->
getModelConfig
();
return
GradientMachine
::
createFromPaddleModelPtr
(
confPtr
,
mode
,
types
);
}
void
GradientMachine
::
forward
(
const
Arguments
&
inArgs
,
Arguments
*
outArgs
,
void
GradientMachine
::
forward
(
const
Arguments
&
inArgs
,
Arguments
*
outArgs
,
PassType
passType
)
{
auto
&
in
=
m
->
cast
<
std
::
vector
<
paddle
::
Argument
>>
(
inArgs
.
getInternalArgumentsPtr
());
...
...
@@ -99,7 +102,8 @@ void GradientMachine::backward(const UpdateCallback& callback) {
}
void
GradientMachine
::
forwardBackward
(
const
Arguments
&
inArgs
,
Arguments
*
outArgs
,
PassType
passType
,
Arguments
*
outArgs
,
PassType
passType
,
const
UpdateCallback
&
callback
)
{
auto
&
in
=
m
->
cast
<
std
::
vector
<
paddle
::
Argument
>>
(
inArgs
.
getInternalArgumentsPtr
());
...
...
@@ -129,7 +133,7 @@ Parameter* GradientMachine::getParameter(size_t i) throw(RangeError) {
void
GradientMachine
::
randParameters
()
{
m
->
machine
->
randParameters
();
}
Matrix
*
GradientMachine
::
getLayerOutput
(
const
std
::
string
&
layerName
)
const
throw
(
UnsupportError
)
{
throw
(
UnsupportError
)
{
auto
nn
=
std
::
dynamic_pointer_cast
<
paddle
::
NeuralNetwork
>
(
m
->
machine
);
if
(
nn
)
{
auto
mat
=
nn
->
getLayerOutput
(
layerName
);
...
...
@@ -140,8 +144,11 @@ Matrix* GradientMachine::getLayerOutput(const std::string& layerName) const
}
SequenceGenerator
*
GradientMachine
::
asSequenceGenerator
(
const
std
::
vector
<
std
::
string
>&
dict
,
size_t
begin_id
,
size_t
end_id
,
size_t
max_length
,
size_t
beam_size
)
{
const
std
::
vector
<
std
::
string
>&
dict
,
size_t
begin_id
,
size_t
end_id
,
size_t
max_length
,
size_t
beam_size
)
{
SequenceGenerator
*
r
=
SequenceGenerator
::
createByGradientMachineSharedPtr
(
&
m
->
machine
);
r
->
setDict
(
dict
);
...
...
paddle/api/Internal.h
浏览文件 @
80c68d38
...
...
@@ -12,7 +12,6 @@ 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. */
#pragma once
#include "PaddleAPI.h"
...
...
@@ -23,7 +22,8 @@ limitations under the License. */
template
<
typename
T1
,
typename
T2
>
void
staticCastVector
(
std
::
vector
<
T2
>*
dest
,
const
std
::
vector
<
T1
>&
src
)
{
dest
->
resize
(
src
.
size
());
std
::
transform
(
src
.
begin
(),
src
.
end
(),
dest
->
begin
(),
[](
T1
t
){
return
static_cast
<
T2
>
(
t
);
});
std
::
transform
(
src
.
begin
(),
src
.
end
(),
dest
->
begin
(),
[](
T1
t
)
{
return
static_cast
<
T2
>
(
t
);
});
}
paddle/api/Matrix.cpp
浏览文件 @
80c68d38
...
...
@@ -12,7 +12,6 @@ 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. */
#include "PaddleAPI.h"
#include "paddle/math/Matrix.h"
#include "paddle/math/SparseMatrix.h"
...
...
@@ -44,17 +43,21 @@ Matrix* Matrix::createZero(size_t height, size_t width, bool useGpu) {
return
m
;
}
Matrix
*
Matrix
::
createDense
(
const
std
::
vector
<
float
>&
data
,
size_t
height
,
size_t
width
,
bool
useGpu
)
{
Matrix
*
Matrix
::
createDense
(
const
std
::
vector
<
float
>&
data
,
size_t
height
,
size_t
width
,
bool
useGpu
)
{
auto
m
=
new
Matrix
();
m
->
m
->
mat
=
paddle
::
Matrix
::
create
(
height
,
width
,
useGpu
);
m
->
m
->
mat
->
copyFrom
(
data
.
data
(),
data
.
size
());
return
m
;
}
Matrix
*
Matrix
::
createDenseFromNumpy
(
float
*
data
,
int
dim1
,
int
dim2
,
bool
copy
,
bool
useGpu
)
throw
(
UnsupportError
)
{
Matrix
*
Matrix
::
createDenseFromNumpy
(
float
*
data
,
int
dim1
,
int
dim2
,
bool
copy
,
bool
useGpu
)
throw
(
UnsupportError
)
{
if
(
useGpu
)
{
/// Gpu mode only supports copy=True
if
(
!
copy
)
{
...
...
@@ -66,7 +69,9 @@ Matrix* Matrix::createDenseFromNumpy(float* data, int dim1, int dim2,
}
}
Matrix
*
Matrix
::
createCpuDenseFromNumpy
(
float
*
data
,
int
dim1
,
int
dim2
,
Matrix
*
Matrix
::
createCpuDenseFromNumpy
(
float
*
data
,
int
dim1
,
int
dim2
,
bool
copy
)
{
auto
m
=
new
Matrix
();
if
(
copy
)
{
...
...
@@ -85,12 +90,20 @@ Matrix* Matrix::createGpuDenseFromNumpy(float* data, int dim1, int dim2) {
return
m
;
}
Matrix
*
Matrix
::
createSparse
(
size_t
height
,
size_t
width
,
size_t
nnz
,
bool
isNonVal
,
bool
isTrans
,
bool
useGpu
)
{
Matrix
*
Matrix
::
createSparse
(
size_t
height
,
size_t
width
,
size_t
nnz
,
bool
isNonVal
,
bool
isTrans
,
bool
useGpu
)
{
auto
m
=
new
Matrix
();
m
->
m
->
mat
=
paddle
::
Matrix
::
createSparseMatrix
(
height
,
width
,
nnz
,
isNonVal
?
paddle
::
NO_VALUE
:
paddle
::
FLOAT_VALUE
,
isTrans
,
useGpu
);
height
,
width
,
nnz
,
isNonVal
?
paddle
::
NO_VALUE
:
paddle
::
FLOAT_VALUE
,
isTrans
,
useGpu
);
return
m
;
}
...
...
@@ -221,7 +234,8 @@ FloatArray Matrix::getData() const {
}
void
Matrix
::
sparseCopyFrom
(
const
std
::
vector
<
int
>&
rows
,
const
std
::
vector
<
int
>&
cols
,
const
std
::
vector
<
int
>&
rows
,
const
std
::
vector
<
int
>&
cols
,
const
std
::
vector
<
float
>&
vals
)
throw
(
UnsupportError
)
{
auto
cpuSparseMat
=
std
::
dynamic_pointer_cast
<
paddle
::
CpuSparseMatrix
>
(
m
->
mat
);
...
...
@@ -240,7 +254,8 @@ void Matrix::sparseCopyFrom(
void
*
Matrix
::
getSharedPtr
()
const
{
return
&
m
->
mat
;
}
void
Matrix
::
toNumpyMatInplace
(
float
**
view_data
,
int
*
dim1
,
void
Matrix
::
toNumpyMatInplace
(
float
**
view_data
,
int
*
dim1
,
int
*
dim2
)
throw
(
UnsupportError
)
{
auto
cpuMat
=
std
::
dynamic_pointer_cast
<
paddle
::
CpuMatrix
>
(
m
->
mat
);
if
(
cpuMat
)
{
...
...
@@ -251,7 +266,8 @@ void Matrix::toNumpyMatInplace(float** view_data, int* dim1,
throw
UnsupportError
();
}
}
void
Matrix
::
copyToNumpyMat
(
float
**
view_m_data
,
int
*
dim1
,
void
Matrix
::
copyToNumpyMat
(
float
**
view_m_data
,
int
*
dim1
,
int
*
dim2
)
throw
(
UnsupportError
)
{
static_assert
(
sizeof
(
paddle
::
real
)
==
sizeof
(
float
),
"Currently PaddleAPI only support for single "
...
...
@@ -269,8 +285,8 @@ void Matrix::copyToNumpyMat(float** view_m_data, int* dim1,
}
else
if
(
auto
gpuMat
=
dynamic_cast
<
paddle
::
GpuMatrix
*>
(
m
->
mat
.
get
()))
{
auto
src
=
gpuMat
->
getData
();
auto
dest
=
*
view_m_data
;
hl_memcpy_device2host
(
dest
,
src
,
sizeof
(
paddle
::
real
)
*
(
*
dim1
)
*
(
*
dim2
));
hl_memcpy_device2host
(
dest
,
src
,
sizeof
(
paddle
::
real
)
*
(
*
dim1
)
*
(
*
dim2
));
}
else
{
LOG
(
WARNING
)
<<
"Unexpected Situation"
;
throw
UnsupportError
();
...
...
@@ -278,7 +294,8 @@ void Matrix::copyToNumpyMat(float** view_m_data, int* dim1,
}
}
void
Matrix
::
copyFromNumpyMat
(
float
*
data
,
int
dim1
,
void
Matrix
::
copyFromNumpyMat
(
float
*
data
,
int
dim1
,
int
dim2
)
throw
(
UnsupportError
,
RangeError
)
{
if
(
isSparse
())
{
throw
UnsupportError
();
...
...
paddle/api/PaddleAPI.h
浏览文件 @
80c68d38
...
...
@@ -12,7 +12,6 @@ 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. */
#pragma once
#include <stddef.h>
...
...
@@ -61,8 +60,8 @@ class RangeError {};
/// Not support Error, such as access GPU memory directly, etc.
class
UnsupportError
:
public
std
::
runtime_error
{
public:
UnsupportError
()
:
std
::
runtime_error
(
" "
)
{};
UnsupportError
(
const
std
::
string
&
message
)
:
std
::
runtime_error
(
message
)
{};
UnsupportError
()
:
std
::
runtime_error
(
" "
){};
UnsupportError
(
const
std
::
string
&
message
)
:
std
::
runtime_error
(
message
){};
};
/// This type will map to python's list of float.
...
...
@@ -112,7 +111,8 @@ public:
/**
* Create A Matrix with height,width, which is filled by zero.
*/
static
Matrix
*
createZero
(
size_t
height
,
size_t
width
,
static
Matrix
*
createZero
(
size_t
height
,
size_t
width
,
bool
useGpu
=
isUsingGpu
());
/**
...
...
@@ -124,8 +124,11 @@ public:
*
* @note the default sparse type is SPARSE_CSR.
*/
static
Matrix
*
createSparse
(
size_t
height
,
size_t
width
,
size_t
nnz
,
bool
isNonVal
=
true
,
bool
trans
=
false
,
static
Matrix
*
createSparse
(
size_t
height
,
size_t
width
,
size_t
nnz
,
bool
isNonVal
=
true
,
bool
trans
=
false
,
bool
useGpu
=
isUsingGpu
());
/**
...
...
@@ -134,13 +137,17 @@ public:
* @param data list of float should be passed in python.
* @note the value will be copy into a new matrix.
*/
static
Matrix
*
createDense
(
const
std
::
vector
<
float
>&
data
,
size_t
height
,
size_t
width
,
bool
useGpu
=
isUsingGpu
());
static
Matrix
*
createDenseFromNumpy
(
float
*
data
,
int
dim1
,
int
dim2
,
bool
copy
=
true
,
bool
useGpu
=
isUsingGpu
())
throw
(
UnsupportError
);
static
Matrix
*
createDense
(
const
std
::
vector
<
float
>&
data
,
size_t
height
,
size_t
width
,
bool
useGpu
=
isUsingGpu
());
static
Matrix
*
createDenseFromNumpy
(
float
*
data
,
int
dim1
,
int
dim2
,
bool
copy
=
true
,
bool
useGpu
=
isUsingGpu
())
throw
(
UnsupportError
);
/**
* Create Cpu Dense Matrix from numpy matrix, dtype=float32
...
...
@@ -151,7 +158,9 @@ public:
* @param copy true if copy into a new matrix, false will create
* matrix inplace.
*/
static
Matrix
*
createCpuDenseFromNumpy
(
float
*
data
,
int
dim1
,
int
dim2
,
static
Matrix
*
createCpuDenseFromNumpy
(
float
*
data
,
int
dim1
,
int
dim2
,
bool
copy
=
false
);
/// Create Gpu Dense Matrix from numpy matrix, dtype=float32
...
...
@@ -171,11 +180,13 @@ public:
* numpy_mat = m.toNumpyMat()
* @endcode
*/
void
toNumpyMatInplace
(
float
**
view_data
,
int
*
dim1
,
void
toNumpyMatInplace
(
float
**
view_data
,
int
*
dim1
,
int
*
dim2
)
throw
(
UnsupportError
);
/// Copy To numpy mat.
void
copyToNumpyMat
(
float
**
view_m_data
,
int
*
dim1
,
void
copyToNumpyMat
(
float
**
view_m_data
,
int
*
dim1
,
int
*
dim2
)
throw
(
UnsupportError
);
/// Copy From Numpy Mat
...
...
@@ -248,15 +259,18 @@ public:
static
Vector
*
create
(
const
std
::
vector
<
float
>&
data
,
bool
useGpu
=
isUsingGpu
());
static
Vector
*
createVectorFromNumpy
(
float
*
data
,
int
dim
,
bool
copy
=
true
,
bool
useGpu
=
isUsingGpu
())
throw
(
UnsupportError
);
static
Vector
*
createVectorFromNumpy
(
float
*
data
,
int
dim
,
bool
copy
=
true
,
bool
useGpu
=
isUsingGpu
())
throw
(
UnsupportError
);
/**
* Create Cpu Vector from numpy array, which dtype=float32
*
* If copy is false, it will create vector inplace.
*/
static
Vector
*
createCpuVectorFromNumpy
(
float
*
data
,
int
dim
,
static
Vector
*
createCpuVectorFromNumpy
(
float
*
data
,
int
dim
,
bool
copy
=
false
);
/// Create Gpu Vector from numpy array, which dtype=float32
...
...
@@ -312,16 +326,19 @@ public:
static
IVector
*
create
(
const
std
::
vector
<
int
>&
data
,
bool
useGpu
=
isUsingGpu
());
static
IVector
*
createVectorFromNumpy
(
int
*
data
,
int
dim
,
bool
copy
=
true
,
bool
useGpu
=
isUsingGpu
())
throw
(
UnsupportError
);
static
IVector
*
createVectorFromNumpy
(
int
*
data
,
int
dim
,
bool
copy
=
true
,
bool
useGpu
=
isUsingGpu
())
throw
(
UnsupportError
);
/**
* Create Cpu IVector from numpy array, which dtype=int32
*
* If copy is false, it will create vector inplace
*/
static
IVector
*
createCpuVectorFromNumpy
(
int
*
data
,
int
dim
,
static
IVector
*
createCpuVectorFromNumpy
(
int
*
data
,
int
dim
,
bool
copy
=
false
);
/**
* Create Gpu IVector from numpy array, which dtype=int32
...
...
@@ -605,7 +622,8 @@ class ParameterTraverseCallback {
public:
~
ParameterTraverseCallback
();
void
apply
(
const
std
::
vector
<
Vector
*>&
vecs
,
const
ParameterConfig
&
config
,
void
apply
(
const
std
::
vector
<
Vector
*>&
vecs
,
const
ParameterConfig
&
config
,
size_t
sparseId
);
private:
...
...
@@ -638,7 +656,8 @@ public:
void
finishBatch
();
void
update
(
const
std
::
vector
<
Vector
*>&
vecs
,
const
ParameterConfig
&
conf
,
void
update
(
const
std
::
vector
<
Vector
*>&
vecs
,
const
ParameterConfig
&
conf
,
size_t
sparseId
=
NO_SPARSE_ID
);
std
::
vector
<
int
>
getParameterTypes
()
const
;
...
...
@@ -678,7 +697,8 @@ public:
* model config by TrainerConfig
*/
static
GradientMachine
*
createByModelConfig
(
ModelConfig
*
conf
,
GradientMatchineCreateMode
mode
=
CREATE_MODE_NORMAL
,
ModelConfig
*
conf
,
GradientMatchineCreateMode
mode
=
CREATE_MODE_NORMAL
,
const
std
::
vector
<
int
>&
parameterTypes
=
defaultParamTypes
);
/**
...
...
@@ -701,7 +721,8 @@ public:
/**
* Combine forward/backward
*/
void
forwardBackward
(
const
Arguments
&
inArgs
,
Arguments
*
outArgs
,
void
forwardBackward
(
const
Arguments
&
inArgs
,
Arguments
*
outArgs
,
PassType
passType
,
const
UpdateCallback
&
callback
=
UpdateCallback
());
...
...
@@ -722,14 +743,17 @@ public:
*/
SequenceGenerator
*
asSequenceGenerator
(
const
std
::
vector
<
std
::
string
>&
dict
=
std
::
vector
<
std
::
string
>
(),
size_t
begin_id
=
0UL
,
size_t
end_id
=
0UL
,
size_t
max_length
=
100UL
,
size_t
begin_id
=
0UL
,
size_t
end_id
=
0UL
,
size_t
max_length
=
100UL
,
size_t
beam_size
=
-
1UL
);
private:
GradientMachinePrivate
*
m
;
static
GradientMachine
*
createFromPaddleModelPtr
(
const
void
*
confPtr
,
GradientMatchineCreateMode
mode
,
const
void
*
confPtr
,
GradientMatchineCreateMode
mode
,
const
std
::
vector
<
int
>&
types
);
// Not to use c++ 11 init-list, so we use static var as function default arg.
...
...
@@ -751,8 +775,8 @@ public:
/// Create A Trainer By TrainerConfig. using paddle command line.
static
Trainer
*
createByCommandLine
()
throw
(
IOError
);
static
Trainer
*
create
(
TrainerConfig
*
optConfig
,
GradientMachine
*
gm
)
throw
(
IOError
);
static
Trainer
*
create
(
TrainerConfig
*
optConfig
,
GradientMachine
*
gm
)
throw
(
IOError
);
/// Start training
void
startTrain
();
...
...
paddle/api/Parameter.cpp
浏览文件 @
80c68d38
...
...
@@ -12,7 +12,6 @@ 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. */
#include "PaddleAPI.h"
#include "paddle/parameter/Parameter.h"
...
...
paddle/api/ParameterOptimizer.cpp
浏览文件 @
80c68d38
...
...
@@ -12,7 +12,6 @@ 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. */
#include "PaddleAPI.h"
#include "PaddleAPIPrivate.h"
#include "paddle/parameter/ParameterOptimizer.h"
...
...
@@ -32,17 +31,21 @@ struct ParameterTraverseCallbackPrivate {
const
paddle
::
ParameterOptimizer
::
TraverseCallback
&
callback
)
:
callback
(
callback
)
{}
void
apply
(
const
std
::
vector
<
Vector
*>&
vecs
,
const
ParameterConfig
&
conf
,
void
apply
(
const
std
::
vector
<
Vector
*>&
vecs
,
const
ParameterConfig
&
conf
,
size_t
sparseId
)
{
std
::
vector
<
paddle
::
VectorPtr
>
real_vecs
;
real_vecs
.
resize
(
vecs
.
size
());
std
::
transform
(
vecs
.
begin
(),
vecs
.
end
(),
real_vecs
.
begin
(),
[](
Vector
*
v
)
{
if
(
v
)
{
return
*
(
paddle
::
VectorPtr
*
)(
v
->
getSharedPtr
());
}
else
{
return
paddle
::
VectorPtr
();
}
});
std
::
transform
(
vecs
.
begin
(),
vecs
.
end
(),
real_vecs
.
begin
(),
[](
Vector
*
v
)
{
if
(
v
)
{
return
*
(
paddle
::
VectorPtr
*
)(
v
->
getSharedPtr
());
}
else
{
return
paddle
::
VectorPtr
();
}
});
paddle
::
ParameterConfig
&
real_conf
=
*
(
paddle
::
ParameterConfig
*
)(
const_cast
<
ParameterConfig
&>
(
conf
)
...
...
@@ -86,10 +89,12 @@ void ParameterOptimizer::startBatch(size_t numSamplesProcessed) {
void
ParameterOptimizer
::
finishBatch
()
{
m
->
optimizer
->
finishBatch
();
}
void
ParameterOptimizer
::
update
(
const
std
::
vector
<
Vector
*>&
vecs
,
const
ParameterConfig
&
conf
,
size_t
sparseId
)
{
ParameterTraverseCallbackPrivate
invoker
([
&
](
const
paddle
::
VectorPtr
_vecs
[],
const
paddle
::
ParameterConfig
&
config
,
size_t
sid
=
-
1UL
)
{
m
->
optimizer
->
update
(
_vecs
,
config
,
sid
);
});
const
ParameterConfig
&
conf
,
size_t
sparseId
)
{
ParameterTraverseCallbackPrivate
invoker
(
[
&
](
const
paddle
::
VectorPtr
_vecs
[],
const
paddle
::
ParameterConfig
&
config
,
size_t
sid
=
-
1UL
)
{
m
->
optimizer
->
update
(
_vecs
,
config
,
sid
);
});
invoker
.
apply
(
vecs
,
conf
,
sparseId
);
}
...
...
@@ -116,8 +121,9 @@ void ParameterTraverseCallback::apply(const std::vector<Vector*>& vecs,
ParameterTraverseCallback
*
ParameterOptimizer
::
needSpecialTraversal
(
const
ParameterConfig
&
config
)
const
{
auto
&
param_config
=
*
(
paddle
::
ParameterConfig
*
)
const_cast
<
ParameterConfig
&>
(
config
).
getRawPtr
();
auto
&
param_config
=
*
(
paddle
::
ParameterConfig
*
)
const_cast
<
ParameterConfig
&>
(
config
)
.
getRawPtr
();
auto
callback
=
m
->
optimizer
->
needSpecialTraversal
(
param_config
);
if
(
callback
)
{
auto
retCallback
=
new
ParameterTraverseCallback
();
...
...
paddle/api/SequenceGenerator.cpp
浏览文件 @
80c68d38
...
...
@@ -12,7 +12,6 @@ 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. */
#include "PaddleAPI.h"
#include "paddle/gserver/gradientmachines/GradientMachine.h"
#include "paddle/parameter/Argument.h"
...
...
@@ -42,8 +41,10 @@ struct Path {
// position
static
void
findNBest
(
paddle
::
GradientMachine
*
gradMachine
,
std
::
vector
<
paddle
::
Argument
>&
inArgs
,
std
::
vector
<
Path
>&
finalPaths
,
size_t
bos_id
,
size_t
eos_id
,
size_t
max_length
)
{
std
::
vector
<
Path
>&
finalPaths
,
size_t
bos_id
,
size_t
eos_id
,
size_t
max_length
)
{
std
::
vector
<
Path
>
paths
;
Path
emptyPath
;
paths
.
push_back
(
emptyPath
);
...
...
@@ -166,7 +167,8 @@ public:
if
(
id
<
getSize
())
{
Path
&
p
=
(
*
path_
)[
id
];
std
::
ostringstream
sout
;
std
::
transform
(
p
.
ids
.
begin
(),
p
.
ids
.
end
(),
std
::
transform
(
p
.
ids
.
begin
(),
p
.
ids
.
end
(),
std
::
ostream_iterator
<
std
::
string
>
(
sout
,
split
?
" "
:
""
),
[
&
](
int
id
)
{
return
(
*
dict_
)[
id
];
});
return
sout
.
str
();
...
...
paddle/api/Trainer.cpp
浏览文件 @
80c68d38
...
...
@@ -64,12 +64,11 @@ Trainer* Trainer::createByCommandLine() throw(IOError) {
Trainer
::
Trainer
(
TrainerConfig
*
config
,
GradientMachine
*
gm
)
:
m
(
new
TrainerPrivate
())
{
m
->
init
(
config
->
m
->
conf
,
/* testing= */
false
,
gm
?
gm
->
m
->
machine
:
nullptr
);
m
->
init
(
config
->
m
->
conf
,
/* testing= */
false
,
gm
?
gm
->
m
->
machine
:
nullptr
);
}
Trainer
*
Trainer
::
create
(
TrainerConfig
*
config
,
GradientMachine
*
gm
)
throw
(
IOError
)
{
Trainer
*
Trainer
::
create
(
TrainerConfig
*
config
,
GradientMachine
*
gm
)
throw
(
IOError
)
{
auto
retv
=
new
Trainer
(
config
,
gm
);
if
(
retv
->
m
->
getConfig
().
IsInitialized
())
{
return
retv
;
...
...
@@ -134,15 +133,17 @@ void Trainer::finishTestPeriod() { m->finishTestPeriod(); }
Matrix
*
Trainer
::
getLayerOutput
(
const
std
::
string
&
layerName
)
{
auto
nn
=
std
::
dynamic_pointer_cast
<
paddle
::
NeuralNetwork
>
(
this
->
m
->
getGradientMachine
());
this
->
m
->
getGradientMachine
());
CHECK
(
nn
)
<<
"trainerInternal_.getGradientMachine() is not NeuralNetwork"
;
auto
m
=
nn
->
getLayerOutput
(
layerName
);
return
Matrix
::
createByPaddleMatrixPtr
(
&
m
);
}
void
Trainer
::
forwardOneBatch
(
size_t
batchSize
)
{
m
->
forwardOneBatch
(
batchSize
);
}
void
Trainer
::
forwardOneBatch
(
size_t
batchSize
)
{
m
->
forwardOneBatch
(
batchSize
);
}
bool
TrainerPrivate
::
forwardOneBatch
(
size_t
batchSize
)
{
bool
TrainerPrivate
::
forwardOneBatch
(
size_t
batchSize
)
{
CHECK
(
dataProvider_
)
<<
"data_provider is not specified"
;
paddle
::
DataBatch
dataBatch
;
int
num
=
dataProvider_
->
getNextBatch
(
batchSize
,
&
dataBatch
);
...
...
@@ -156,7 +157,6 @@ bool TrainerPrivate::forwardOneBatch(size_t batchSize) {
void
TrainerPrivate
::
forwardOneDataBatch
(
const
std
::
vector
<
paddle
::
Argument
>&
inArgs
)
{
std
::
vector
<
paddle
::
Argument
>&
outArgs
=
forwardOutput_
;
if
(
config_
->
getOptConfig
().
use_sparse_remote_updater
())
{
...
...
paddle/api/Util.cpp
浏览文件 @
80c68d38
...
...
@@ -37,13 +37,15 @@ FloatArray::FloatArray(const float* b, const size_t l)
IntArray
::
IntArray
(
const
int
*
b
,
const
size_t
l
,
bool
f
)
:
buf
(
b
),
length
(
l
),
needFree
(
f
)
{}
IntWithFloatArray
::
IntWithFloatArray
(
const
float
*
v
,
const
int
*
i
,
size_t
l
,
IntWithFloatArray
::
IntWithFloatArray
(
const
float
*
v
,
const
int
*
i
,
size_t
l
,
bool
f
)
:
valBuf
(
v
),
idxBuf
(
i
),
length
(
l
),
needFree
(
f
)
{}
bool
isUsingGpu
()
{
return
FLAGS_use_gpu
;
}
bool
isUsingGpu
()
{
return
FLAGS_use_gpu
;
}
void
setUseGpu
(
bool
useGpu
)
{
FLAGS_use_gpu
=
useGpu
;
}
void
setUseGpu
(
bool
useGpu
)
{
FLAGS_use_gpu
=
useGpu
;
}
bool
isGpuVersion
()
{
#ifdef PADDLE_ONLY_CPU
...
...
paddle/api/Vector.cpp
浏览文件 @
80c68d38
...
...
@@ -12,7 +12,6 @@ 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. */
#include "PaddleAPI.h"
#include "paddle/math/Vector.h"
...
...
@@ -39,8 +38,10 @@ IVector* IVector::create(const std::vector<int>& data, bool useGpu) {
return
v
;
}
IVector
*
IVector
::
createVectorFromNumpy
(
int
*
data
,
int
dim
,
bool
copy
,
bool
useGpu
)
throw
(
UnsupportError
){
IVector
*
IVector
::
createVectorFromNumpy
(
int
*
data
,
int
dim
,
bool
copy
,
bool
useGpu
)
throw
(
UnsupportError
)
{
if
(
useGpu
)
{
/// if use gpu only copy=true is supported
if
(
!
copy
)
{
...
...
@@ -137,8 +138,8 @@ void IVector::copyToNumpyArray(int** view_m_data, int* dim1) {
if
(
auto
cpuVec
=
dynamic_cast
<
paddle
::
CpuIVector
*>
(
m
->
vec
.
get
()))
{
std
::
memcpy
(
*
view_m_data
,
cpuVec
->
getData
(),
sizeof
(
int
)
*
(
*
dim1
));
}
else
if
(
auto
gpuVec
=
dynamic_cast
<
paddle
::
GpuIVector
*>
(
m
->
vec
.
get
()))
{
hl_memcpy_device2host
(
*
view_m_data
,
gpuVec
->
getData
(),
sizeof
(
int
)
*
(
*
dim1
));
hl_memcpy_device2host
(
*
view_m_data
,
gpuVec
->
getData
(),
sizeof
(
int
)
*
(
*
dim1
));
}
else
{
LOG
(
INFO
)
<<
"Unexpected situation"
;
}
...
...
@@ -201,8 +202,10 @@ Vector* Vector::createByPaddleVectorPtr(void* ptr) {
}
}
Vector
*
Vector
::
createVectorFromNumpy
(
float
*
data
,
int
dim
,
bool
copy
,
bool
useGpu
)
throw
(
UnsupportError
){
Vector
*
Vector
::
createVectorFromNumpy
(
float
*
data
,
int
dim
,
bool
copy
,
bool
useGpu
)
throw
(
UnsupportError
)
{
if
(
useGpu
)
{
/// if use gpu only copy=True is supported
if
(
!
copy
)
{
...
...
@@ -251,8 +254,8 @@ void Vector::copyToNumpyArray(float** view_m_data, int* dim1) {
if
(
auto
cpuVec
=
dynamic_cast
<
paddle
::
CpuVector
*>
(
m
->
vec
.
get
()))
{
std
::
memcpy
(
*
view_m_data
,
cpuVec
->
getData
(),
sizeof
(
float
)
*
(
*
dim1
));
}
else
if
(
auto
gpuVec
=
dynamic_cast
<
paddle
::
CpuVector
*>
(
m
->
vec
.
get
()))
{
hl_memcpy_device2host
(
*
view_m_data
,
gpuVec
->
getData
(),
sizeof
(
float
)
*
(
*
dim1
));
hl_memcpy_device2host
(
*
view_m_data
,
gpuVec
->
getData
(),
sizeof
(
float
)
*
(
*
dim1
));
}
else
{
LOG
(
INFO
)
<<
"Unexpected situation"
;
}
...
...
paddle/cuda/include/hl_activation_functions.h
浏览文件 @
80c68d38
...
...
@@ -12,7 +12,6 @@ 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. */
#ifndef HL_ACTIVATION_FUNCTIONS_H_
#define HL_ACTIVATION_FUNCTIONS_H_
...
...
@@ -21,11 +20,8 @@ limitations under the License. */
/**
* Active functions: sigmoid, relu, tanh and linear.
*/
#define HPPL_ACTIVE_FUNCTION {hppl::sigmoid, \
hppl::relu, \
hppl::tanh, \
hppl::linear \
}
#define HPPL_ACTIVE_FUNCTION \
{ hppl::sigmoid, hppl::relu, hppl::tanh, hppl::linear }
namespace
hppl
{
...
...
@@ -42,18 +38,18 @@ public:
#ifdef __NVCC__
namespace
gpu
{
static
__device__
Active
<
real
>::
forward
forward
[]
=
HPPL_ACTIVE_FUNCTION
;
static
__device__
Active
<
real
>::
forward
forward
[]
=
HPPL_ACTIVE_FUNCTION
;
static
__device__
Active
<
real
>::
backward
backward
[]
=
HPPL_ACTIVE_FUNCTION
;
}
#else
namespace
cpu
{
static
Active
<
real
>::
forward
forward
[]
=
HPPL_ACTIVE_FUNCTION
;
static
Active
<
real
>::
forward
forward
[]
=
HPPL_ACTIVE_FUNCTION
;
static
Active
<
real
>::
backward
backward
[]
=
HPPL_ACTIVE_FUNCTION
;
}
#ifdef __AVX__
namespace
avx
{
static
Active
<
__m256
>::
forward
forward
[]
=
HPPL_ACTIVE_FUNCTION
;
static
Active
<
__m256
>::
forward
forward
[]
=
HPPL_ACTIVE_FUNCTION
;
static
Active
<
__m256
>::
backward
backward
[]
=
HPPL_ACTIVE_FUNCTION
;
}
#endif
...
...
paddle/cuda/include/hl_aggregate.h
浏览文件 @
80c68d38
...
...
@@ -12,7 +12,6 @@ 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. */
#ifndef HL_AGGREGATE_H_
#define HL_AGGREGATE_H_
...
...
paddle/cuda/include/hl_avx_functions.h
浏览文件 @
80c68d38
...
...
@@ -12,22 +12,21 @@ 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. */
#ifndef HL_AVX_FUNCTIONS_H_
#define HL_AVX_FUNCTIONS_H_
#include <immintrin.h>
namespace
hppl
{
__m256
relu
(
const
__m256
a
);
__m256
sigmoid
(
const
__m256
a
);
__m256
tanh
(
const
__m256
a
);
__m256
linear
(
const
__m256
a
);
__m256
relu
(
const
__m256
a
,
const
__m256
b
);
__m256
sigmoid
(
const
__m256
a
,
const
__m256
b
);
__m256
tanh
(
const
__m256
a
,
const
__m256
b
);
__m256
linear
(
const
__m256
a
,
const
__m256
b
);
__m256
relu
(
const
__m256
a
);
__m256
sigmoid
(
const
__m256
a
);
__m256
tanh
(
const
__m256
a
);
__m256
linear
(
const
__m256
a
);
__m256
relu
(
const
__m256
a
,
const
__m256
b
);
__m256
sigmoid
(
const
__m256
a
,
const
__m256
b
);
__m256
tanh
(
const
__m256
a
,
const
__m256
b
);
__m256
linear
(
const
__m256
a
,
const
__m256
b
);
}
// namespace hppl
#endif // HL_AVX_FUNCTIONS_H_
paddle/cuda/include/hl_base.h
浏览文件 @
80c68d38
...
...
@@ -12,8 +12,6 @@ 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. */
#ifndef HL_BASE_H_
#define HL_BASE_H_
...
...
@@ -33,36 +31,36 @@ limitations under the License. */
* HPPL_STREAM_DEFAULT is HPPL default stream.
*/
typedef
enum
{
HPPL_STREAM_DEFAULT
=
0
,
/* Thread Default Stream*/
HPPL_STREAM_1
=
1
,
HPPL_STREAM_2
=
2
,
HPPL_STREAM_3
=
3
,
HPPL_STREAM_4
=
4
,
HPPL_THREAD_STREAM_1
=
5
,
HPPL_THREAD_STREAM_2
=
6
,
HPPL_THREAD_STREAM_3
=
7
,
HPPL_THREAD_STREAM_4
=
8
,
HPPL_STREAM_END
HPPL_STREAM_DEFAULT
=
0
,
/* Thread Default Stream*/
HPPL_STREAM_1
=
1
,
HPPL_STREAM_2
=
2
,
HPPL_STREAM_3
=
3
,
HPPL_STREAM_4
=
4
,
HPPL_THREAD_STREAM_1
=
5
,
HPPL_THREAD_STREAM_2
=
6
,
HPPL_THREAD_STREAM_3
=
7
,
HPPL_THREAD_STREAM_4
=
8
,
HPPL_STREAM_END
}
hl_stream_t
;
/**
* @brief HPPL activation mode.
*/
typedef
enum
{
HL_ACTIVATION_SIGMOID
=
0
,
HL_ACTIVATION_RELU
=
1
,
HL_ACTIVATION_TANH
=
2
,
HL_ACTIVATION_LINEAR
=
3
,
HL_ACTIVATION_END
HL_ACTIVATION_SIGMOID
=
0
,
HL_ACTIVATION_RELU
=
1
,
HL_ACTIVATION_TANH
=
2
,
HL_ACTIVATION_LINEAR
=
3
,
HL_ACTIVATION_END
}
hl_activation_mode_t
;
/**
* @brief Transpose type.
*/
typedef
enum
{
HPPL_OP_N
=
0
,
/* transpose */
HPPL_OP_T
=
1
,
/* non transpose */
HPPL_OP_END
HPPL_OP_N
=
0
,
/* transpose */
HPPL_OP_T
=
1
,
/* non transpose */
HPPL_OP_END
}
hl_trans_op_t
;
/**
...
...
@@ -148,23 +146,21 @@ typedef struct {
* @brief Sparse matrix value type.
*/
typedef
enum
{
HL_NO_VALUE
=
0
,
/* matrix values only 0 or 1 */
HL_FLOAT_VALUE
=
1
,
HL_VALUE_END
HL_NO_VALUE
=
0
,
/* matrix values only 0 or 1 */
HL_FLOAT_VALUE
=
1
,
HL_VALUE_END
}
hl_matrix_value_t
;
/**
* @brief HPPL matrix format.
*/
typedef
enum
{
HL_SPARSE_CSR
=
0
,
HL_SPARSE_CSC
=
1
,
HL_SPARSE_END
HL_SPARSE_CSR
=
0
,
HL_SPARSE_CSC
=
1
,
HL_SPARSE_END
}
hl_matrix_format_t
;
typedef
struct
_hl_matrix_s
*
hl_matrix_s
;
typedef
struct
_hl_matrix_s
*
hl_matrix_s
;
/**
* @brief HPPL sparse matrix.
...
...
@@ -177,12 +173,12 @@ typedef struct _hl_matrix_s * hl_matrix_s;
* @param nnz nonzero values of sparse matrix.
*/
typedef
struct
{
hl_matrix_s
matrix
;
hl_matrix_format_t
format
;
hl_matrix_value_t
type
;
int
rows
;
int
cols
;
size_t
nnz
;
hl_matrix_s
matrix
;
hl_matrix_format_t
format
;
hl_matrix_value_t
type
;
int
rows
;
int
cols
;
size_t
nnz
;
}
_hl_sparse_matrix_s
,
*
hl_sparse_matrix_s
;
#ifndef PADDLE_TYPE_DOUBLE
...
...
@@ -195,7 +191,7 @@ typedef struct {
*
* HL_FLOAT_MIN: 1.17549435e-38F
*/
#define HL_FLOAT_MAX
3.40282347e+38F
#define HL_FLOAT_MAX 3.40282347e+38F
/**
* if real == double
*
...
...
@@ -203,20 +199,18 @@ typedef struct {
*
* HL_FLOAT_MIN: 2.2250738585072014e-308
*/
#define HL_FLOAT_MIN
1.17549435e-38F
#define HL_FLOAT_MIN 1.17549435e-38F
#else
#define HL_FLOAT_MAX
1.7976931348623157e+308
#define HL_FLOAT_MIN
2.2250738585072014e-308
#define HL_FLOAT_MAX 1.7976931348623157e+308
#define HL_FLOAT_MIN 2.2250738585072014e-308
#endif
/**
* The maximum input value for exp, used to avoid overflow problem.
*
* Currently only used for tanh function.
*/
#define EXP_MAX_INPUT 40.0
#define EXP_MAX_INPUT 40.0
/**
* @brief DIVUP(x, y) is similar to ceil(x / y).
...
...
@@ -224,7 +218,7 @@ typedef struct {
* the size of blockDim.
*/
#ifndef DIVUP
#define DIVUP(x, y) (((x) + (y)
-
1) / (y))
#define DIVUP(x, y) (((x) + (y)
-
1) / (y))
#endif
#ifdef __NVCC__
...
...
@@ -233,7 +227,7 @@ typedef struct {
#include "hl_cuda.h"
#include "cuda_runtime.h"
extern
__thread
bool
g_sync_flag
;
extern
__thread
bool
g_sync_flag
;
extern
__thread
cudaStream_t
default_stream
;
#define STREAM_DEFAULT default_stream
...
...
@@ -241,16 +235,15 @@ extern __thread cudaStream_t default_stream;
* @brief Check cuda kernel execution.
* @param msg error string
*/
#define CHECK_SYNC(msg) \
if (true == g_sync_flag) { \
hl_stream_synchronize(HPPL_STREAM_DEFAULT); \
cudaError_t err \
= (cudaError_t)hl_get_device_last_error(); \
CHECK_EQ(cudaSuccess, err) << "[" << msg << "] " \
<< "CUDA error: " \
<< hl_get_device_error_string((size_t)err); \
#define CHECK_SYNC(msg) \
if (true == g_sync_flag) { \
hl_stream_synchronize(HPPL_STREAM_DEFAULT); \
cudaError_t err = (cudaError_t)hl_get_device_last_error(); \
CHECK_EQ(cudaSuccess, err) \
<< "[" << msg << "] " \
<< "CUDA error: " << hl_get_device_error_string((size_t)err); \
}
#endif
/* __NVCC__ */
#endif
/* __NVCC__ */
#endif
/* HL_BASE_H_ */
#endif
/* HL_BASE_H_ */
paddle/cuda/include/hl_batch_transpose.h
浏览文件 @
80c68d38
...
...
@@ -12,7 +12,6 @@ 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. */
#ifndef HL_BATCH_TRANSPOSE_H_
#define HL_BATCH_TRANSPOSE_H_
...
...
@@ -31,10 +30,7 @@ limitations under the License. */
* order. Each batch has height * width data, which are
* arranged in height-first (or row-first) manner.
*/
extern
void
batchTranspose
(
const
real
*
input
,
real
*
output
,
int
width
,
int
height
,
int
batchSize
);
extern
void
batchTranspose
(
const
real
*
input
,
real
*
output
,
int
width
,
int
height
,
int
batchSize
);
#endif // HL_BATCH_TRANSPOSE_H_
paddle/cuda/include/hl_cnn.h
浏览文件 @
80c68d38
...
...
@@ -12,7 +12,6 @@ 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. */
#ifndef HL_CNN_H_
#define HL_CNN_H_
...
...
@@ -37,15 +36,21 @@ limitations under the License. */
* @param[in] alpha
* @param[in] beta
*/
extern
void
hl_shrink_col2feature
(
const
real
*
dataCol
,
size_t
channels
,
size_t
height
,
size_t
width
,
size_t
blockH
,
size_t
blockW
,
size_t
strideH
,
size_t
strideW
,
size_t
paddingH
,
size_t
paddingW
,
size_t
outputH
,
size_t
outputW
,
real
*
dataIm
,
real
alpha
=
1
.
0
f
,
real
beta
=
0
.
0
f
);
extern
void
hl_shrink_col2feature
(
const
real
*
dataCol
,
size_t
channels
,
size_t
height
,
size_t
width
,
size_t
blockH
,
size_t
blockW
,
size_t
strideH
,
size_t
strideW
,
size_t
paddingH
,
size_t
paddingW
,
size_t
outputH
,
size_t
outputW
,
real
*
dataIm
,
real
alpha
=
1
.
0
f
,
real
beta
=
0
.
0
f
);
/**
* @brief Expand feature to column.
...
...
@@ -65,14 +70,19 @@ extern void hl_shrink_col2feature(
* @param[out] dataCol expand data.
*
*/
extern
void
hl_expand_feature2col
(
const
real
*
dataIm
,
size_t
channels
,
size_t
height
,
size_t
width
,
size_t
blockH
,
size_t
blockW
,
size_t
strideH
,
size_t
strideW
,
size_t
paddingH
,
size_t
paddingW
,
size_t
outputH
,
size_t
outputW
,
real
*
dataCol
);
extern
void
hl_expand_feature2col
(
const
real
*
dataIm
,
size_t
channels
,
size_t
height
,
size_t
width
,
size_t
blockH
,
size_t
blockW
,
size_t
strideH
,
size_t
strideW
,
size_t
paddingH
,
size_t
paddingW
,
size_t
outputH
,
size_t
outputW
,
real
*
dataCol
);
/**
* @brief Maximum pool forward.
...
...
@@ -94,15 +104,21 @@ extern void hl_expand_feature2col(
* @param[in] tgtStride stride between output data samples.
*
*/
extern
void
hl_maxpool_forward
(
const
int
frameCnt
,
const
real
*
inputData
,
const
int
channels
,
const
int
height
,
const
int
width
,
const
int
pooledH
,
const
int
pooledW
,
const
int
sizeX
,
const
int
sizeY
,
const
int
strideH
,
const
int
strideW
,
const
int
paddingH
,
const
int
paddingW
,
real
*
tgtData
,
const
int
tgtStride
);
extern
void
hl_maxpool_forward
(
const
int
frameCnt
,
const
real
*
inputData
,
const
int
channels
,
const
int
height
,
const
int
width
,
const
int
pooledH
,
const
int
pooledW
,
const
int
sizeX
,
const
int
sizeY
,
const
int
strideH
,
const
int
strideW
,
const
int
paddingH
,
const
int
paddingW
,
real
*
tgtData
,
const
int
tgtStride
);
/**
* @brief Maximum pool backward.
...
...
@@ -125,20 +141,28 @@ extern void hl_maxpool_forward(
* @param[in] paddingH padding height.
* @param[in] paddingW padding width.
* @param[out] targetGrad output grad.
* @param[in] outStride stride between output data samples.
* @param[in] outStride stride between output data samples.
*
*/
extern
void
hl_maxpool_backward
(
const
int
frameCnt
,
const
real
*
inputData
,
const
real
*
outData
,
const
real
*
outGrad
,
const
int
channels
,
const
int
height
,
const
int
width
,
const
int
pooledH
,
const
int
pooledW
,
const
int
sizeX
,
const
int
sizeY
,
const
int
strideH
,
const
int
strideW
,
const
int
paddingH
,
const
int
paddingW
,
real
scaleA
,
real
scaleB
,
real
*
targetGrad
,
const
int
outStride
);
extern
void
hl_maxpool_backward
(
const
int
frameCnt
,
const
real
*
inputData
,
const
real
*
outData
,
const
real
*
outGrad
,
const
int
channels
,
const
int
height
,
const
int
width
,
const
int
pooledH
,
const
int
pooledW
,
const
int
sizeX
,
const
int
sizeY
,
const
int
strideH
,
const
int
strideW
,
const
int
paddingH
,
const
int
paddingW
,
real
scaleA
,
real
scaleB
,
real
*
targetGrad
,
const
int
outStride
);
/**
* @brief Averge pool forward.
...
...
@@ -160,15 +184,21 @@ extern void hl_maxpool_backward(
* @param[in] tgtStride stride between output data samples.
*
*/
extern
void
hl_avgpool_forward
(
const
int
frameCnt
,
const
real
*
inputData
,
const
int
channels
,
const
int
height
,
const
int
width
,
const
int
pooledH
,
const
int
pooledW
,
const
int
sizeX
,
const
int
sizeY
,
const
int
strideH
,
const
int
strideW
,
const
int
paddingH
,
const
int
paddingW
,
real
*
tgtData
,
const
int
tgtStride
);
extern
void
hl_avgpool_forward
(
const
int
frameCnt
,
const
real
*
inputData
,
const
int
channels
,
const
int
height
,
const
int
width
,
const
int
pooledH
,
const
int
pooledW
,
const
int
sizeX
,
const
int
sizeY
,
const
int
strideH
,
const
int
strideW
,
const
int
paddingH
,
const
int
paddingW
,
real
*
tgtData
,
const
int
tgtStride
);
/**
* @brief Maximum pool backward.
...
...
@@ -189,19 +219,26 @@ extern void hl_avgpool_forward(
* @param[in] scaleA scale.
* @param[in] scaleB scale.
* @param[out] backGrad output grad.
* @param[in] outStride stride between output data samples.
* @param[in] outStride stride between output data samples.
*
*/
extern
void
hl_avgpool_backward
(
const
int
frameCnt
,
const
real
*
outGrad
,
const
int
channels
,
const
int
height
,
const
int
width
,
const
int
pooledH
,
const
int
pooledW
,
const
int
sizeX
,
const
int
sizeY
,
const
int
strideH
,
const
int
strideW
,
int
paddingH
,
int
paddingW
,
real
scaleA
,
real
scaleB
,
real
*
backGrad
,
const
int
outStride
);
extern
void
hl_avgpool_backward
(
const
int
frameCnt
,
const
real
*
outGrad
,
const
int
channels
,
const
int
height
,
const
int
width
,
const
int
pooledH
,
const
int
pooledW
,
const
int
sizeX
,
const
int
sizeY
,
const
int
strideH
,
const
int
strideW
,
int
paddingH
,
int
paddingW
,
real
scaleA
,
real
scaleB
,
real
*
backGrad
,
const
int
outStride
);
/**
* @brief Cross-map-respose normalize forward.
...
...
@@ -218,10 +255,16 @@ extern void hl_avgpool_backward(
* @param[in] beta scale.
*
*/
extern
void
hl_CMRNorm_forward
(
size_t
frameCnt
,
const
real
*
in
,
real
*
scale
,
real
*
out
,
size_t
channels
,
size_t
height
,
size_t
width
,
size_t
sizeX
,
real
alpha
,
real
beta
);
extern
void
hl_CMRNorm_forward
(
size_t
frameCnt
,
const
real
*
in
,
real
*
scale
,
real
*
out
,
size_t
channels
,
size_t
height
,
size_t
width
,
size_t
sizeX
,
real
alpha
,
real
beta
);
/**
* @brief Cross-map-respose normalize backward.
...
...
@@ -240,11 +283,18 @@ extern void hl_CMRNorm_forward(
* @param[in] beta scale.
*
*/
extern
void
hl_CMRNorm_backward
(
size_t
frameCnt
,
const
real
*
inV
,
const
real
*
scale
,
const
real
*
outV
,
const
real
*
outDiff
,
real
*
inDiff
,
size_t
channels
,
size_t
height
,
size_t
width
,
size_t
sizeX
,
real
alpha
,
real
beta
);
extern
void
hl_CMRNorm_backward
(
size_t
frameCnt
,
const
real
*
inV
,
const
real
*
scale
,
const
real
*
outV
,
const
real
*
outDiff
,
real
*
inDiff
,
size_t
channels
,
size_t
height
,
size_t
width
,
size_t
sizeX
,
real
alpha
,
real
beta
);
/**
* @brief Bilinear interpolation forward.
...
...
@@ -278,24 +328,24 @@ extern void hl_bilinear_forward(const real* inData,
const
real
ratioH
,
const
real
ratioW
);
/**
* @brief Bilinear interpolation backward.
*
* @param[out] inGrad input gradient.
* @param[in] inImgH input image height.
* @param[in] inImgW input image width.
* @param[in] inputH input batchSize.
* @param[in] inputW input image data dim.
* @param[in] outGrad output gradient.
* @param[in] outImgH output image height.
* @param[in] outImgW output image width.
* @param[in] outputH output batchSize.
* @param[in] outputW output image data dim.
* @param[in] numChannels number of channels.
* @param[in] ratioH inImgH / outImgH.
* @param[in] ratioW inImgW / outImgW.
*
*/
/**
* @brief Bilinear interpolation backward.
*
* @param[out] inGrad input gradient.
* @param[in] inImgH input image height.
* @param[in] inImgW input image width.
* @param[in] inputH input batchSize.
* @param[in] inputW input image data dim.
* @param[in] outGrad output gradient.
* @param[in] outImgH output image height.
* @param[in] outImgW output image width.
* @param[in] outputH output batchSize.
* @param[in] outputW output image data dim.
* @param[in] numChannels number of channels.
* @param[in] ratioH inImgH / outImgH.
* @param[in] ratioW inImgW / outImgW.
*
*/
extern
void
hl_bilinear_backward
(
real
*
inGrad
,
const
size_t
inImgH
,
const
size_t
inImgW
,
...
...
@@ -321,9 +371,13 @@ extern void hl_bilinear_backward(real* inGrad,
* @param[in] featLen feature length = image height * image width.
* @param[in] groups number of groups.
*/
extern
void
hl_maxout_forward
(
const
real
*
inData
,
real
*
outData
,
int
*
idData
,
size_t
batchSize
,
size_t
size
,
size_t
featLen
,
size_t
groups
);
extern
void
hl_maxout_forward
(
const
real
*
inData
,
real
*
outData
,
int
*
idData
,
size_t
batchSize
,
size_t
size
,
size_t
featLen
,
size_t
groups
);
/**
* @brief MaxOut backward.
...
...
@@ -336,8 +390,12 @@ extern void hl_maxout_forward(
* @param[in] featLen feature length = image height * image width.
* @param[in] groups number of groups.
*/
extern
void
hl_maxout_backward
(
real
*
inGrad
,
const
real
*
outGrad
,
const
int
*
idData
,
size_t
batchSize
,
size_t
size
,
size_t
featLen
,
size_t
groups
);
extern
void
hl_maxout_backward
(
real
*
inGrad
,
const
real
*
outGrad
,
const
int
*
idData
,
size_t
batchSize
,
size_t
size
,
size_t
featLen
,
size_t
groups
);
#endif
/* HL_CNN_H_ */
paddle/cuda/include/hl_cuda.h
浏览文件 @
80c68d38
...
...
@@ -12,7 +12,6 @@ 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. */
#ifndef HL_CUDA_H_
#define HL_CUDA_H_
...
...
@@ -22,8 +21,7 @@ limitations under the License. */
/**
* @brief HPPL event.
*/
typedef
struct
_hl_event_st
*
hl_event_t
;
typedef
struct
_hl_event_st
*
hl_event_t
;
/**
* @brief return cuda runtime api version.
...
...
@@ -42,7 +40,7 @@ extern void hl_start();
* if device is NULL, will start all GPU.
* @param[in] number number of devices.
*/
extern
void
hl_specify_devices_start
(
int
*
device
,
int
number
);
extern
void
hl_specify_devices_start
(
int
*
device
,
int
number
);
/**
* @brief Queries if a device may directly access a peer device's memory.
...
...
@@ -126,7 +124,7 @@ extern int hl_get_device();
*
* @return dest_d pointer to device memory.
*/
extern
void
*
hl_malloc_device
(
size_t
size
);
extern
void
*
hl_malloc_device
(
size_t
size
);
/**
* @brief Free device memory.
...
...
@@ -143,7 +141,7 @@ extern void hl_free_mem_device(void *dest_d);
*
* @return dest_h pointer to host memory.
*/
extern
void
*
hl_malloc_host
(
size_t
size
);
extern
void
*
hl_malloc_host
(
size_t
size
);
/**
* @brief Free host page-lock memory.
...
...
@@ -228,9 +226,9 @@ extern void hl_srand(unsigned int seed);
* @param[in] stream stream id.
*/
extern
void
hl_memcpy_async
(
void
*
dst
,
void
*
src
,
size_t
size
,
hl_stream_t
stream
);
void
*
src
,
size_t
size
,
hl_stream_t
stream
);
/**
* @brief Waits for stream tasks to complete.
...
...
@@ -261,8 +259,7 @@ extern void hl_destroy_event(hl_event_t event);
*
* @return time Time between start and end in ms.
*/
extern
float
hl_event_elapsed_time
(
hl_event_t
start
,
hl_event_t
end
);
extern
float
hl_event_elapsed_time
(
hl_event_t
start
,
hl_event_t
end
);
/**
* @brief Records an event.
...
...
@@ -300,7 +297,7 @@ extern void hl_set_device_flags_block();
/**
* @brief Returns the last error string from a cuda runtime call.
*/
extern
const
char
*
hl_get_device_error_string
();
extern
const
char
*
hl_get_device_error_string
();
/**
* @brief Returns the last error string from a cuda runtime call.
...
...
@@ -309,7 +306,7 @@ extern const char* hl_get_device_error_string();
*
* @see hl_get_device_last_error()
*/
extern
const
char
*
hl_get_device_error_string
(
size_t
err
);
extern
const
char
*
hl_get_device_error_string
(
size_t
err
);
/**
* @brief Returns the last error number.
...
...
paddle/cuda/include/hl_cuda_cublas.h
浏览文件 @
80c68d38
...
...
@@ -12,7 +12,6 @@ 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. */
#ifndef HL_CUDA_CUBLAS_H_
#define HL_CUDA_CUBLAS_H_
...
...
@@ -29,12 +28,8 @@ limitations under the License. */
* @param[in] ldc the first dimension of C_d.
*
*/
extern
void
hl_matrix_transpose
(
real
*
A_d
,
real
*
C_d
,
int
dimM
,
int
dimN
,
int
lda
,
int
ldc
);
extern
void
hl_matrix_transpose
(
real
*
A_d
,
real
*
C_d
,
int
dimM
,
int
dimN
,
int
lda
,
int
ldc
);
/*
* @brief Matrix transpose, while lda = dimN, ldc = dimM.
...
...
@@ -45,10 +40,7 @@ extern void hl_matrix_transpose(real *A_d,
* @param[in] dimN matrix width.
*
*/
extern
void
hl_matrix_transpose
(
real
*
A_d
,
real
*
C_d
,
int
dimM
,
int
dimN
);
extern
void
hl_matrix_transpose
(
real
*
A_d
,
real
*
C_d
,
int
dimM
,
int
dimN
);
/*
* @brief Matrix inverse
...
...
@@ -60,11 +52,7 @@ extern void hl_matrix_transpose(real *A_d,
* @param[in] ldc the first dimension of C_d
*
*/
extern
void
hl_matrix_inverse
(
real
*
A_d
,
real
*
C_d
,
int
dimN
,
int
lda
,
int
ldc
);
extern
void
hl_matrix_inverse
(
real
*
A_d
,
real
*
C_d
,
int
dimN
,
int
lda
,
int
ldc
);
/**
* @brief C_d = alpha*(op(A_d) * op(B_d)) + beta*C_d
...
...
@@ -84,12 +72,19 @@ extern void hl_matrix_inverse(real *A_d,
* @param[in] ldc the first dimension of C_d.
*
*/
extern
void
hl_matrix_mul
(
real
*
A_d
,
hl_trans_op_t
transa
,
real
*
B_d
,
hl_trans_op_t
transb
,
extern
void
hl_matrix_mul
(
real
*
A_d
,
hl_trans_op_t
transa
,
real
*
B_d
,
hl_trans_op_t
transb
,
real
*
C_d
,
int
dimM
,
int
dimN
,
int
dimK
,
real
alpha
,
real
beta
,
int
lda
,
int
ldb
,
int
ldc
);
int
dimM
,
int
dimN
,
int
dimK
,
real
alpha
,
real
beta
,
int
lda
,
int
ldb
,
int
ldc
);
/**
* @brief C_d = alpha*(op(A_d) * op(B_d)) + beta*C_d
...
...
@@ -106,11 +101,16 @@ extern void hl_matrix_mul(real *A_d, hl_trans_op_t transa,
* @param[in] beta scalar used for multiplication.
*
*/
extern
void
hl_matrix_mul
(
real
*
A_d
,
hl_trans_op_t
transa
,
real
*
B_d
,
hl_trans_op_t
transb
,
extern
void
hl_matrix_mul
(
real
*
A_d
,
hl_trans_op_t
transa
,
real
*
B_d
,
hl_trans_op_t
transb
,
real
*
C_d
,
int
dimM
,
int
dimN
,
int
dimK
,
real
alpha
,
real
beta
);
int
dimM
,
int
dimN
,
int
dimK
,
real
alpha
,
real
beta
);
/**
* @brief This function performs the matrix-vector multiplication.
...
...
@@ -132,11 +132,17 @@ extern void hl_matrix_mul(real *A_d, hl_trans_op_t transa,
*
*/
extern
void
hl_matrix_mul_vector
(
real
*
A_d
,
hl_trans_op_t
trans
,
real
*
B_d
,
real
*
C_d
,
int
dimM
,
int
dimN
,
real
alpha
,
real
beta
,
int
lda
,
int
incb
,
int
incc
);
extern
void
hl_matrix_mul_vector
(
real
*
A_d
,
hl_trans_op_t
trans
,
real
*
B_d
,
real
*
C_d
,
int
dimM
,
int
dimN
,
real
alpha
,
real
beta
,
int
lda
,
int
incb
,
int
incc
);
/**
* @brief This function performs the matrix-vector multiplication.
...
...
@@ -154,9 +160,13 @@ extern void hl_matrix_mul_vector(real *A_d, hl_trans_op_t trans,
* @param[in] beta scalar used for multiplication.
*
*/
extern
void
hl_matrix_mul_vector
(
real
*
A_d
,
hl_trans_op_t
trans
,
real
*
B_d
,
real
*
C_d
,
int
dimM
,
int
dimN
,
real
alpha
,
real
beta
);
extern
void
hl_matrix_mul_vector
(
real
*
A_d
,
hl_trans_op_t
trans
,
real
*
B_d
,
real
*
C_d
,
int
dimM
,
int
dimN
,
real
alpha
,
real
beta
);
#endif
/* HL_CUDA_CUBLAS_H_ */
paddle/cuda/include/hl_cuda_cudnn.h
浏览文件 @
80c68d38
...
...
@@ -12,7 +12,6 @@ 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. */
#ifndef HL_CUDA_CUDNN_H_
#define HL_CUDA_CUDNN_H_
...
...
@@ -22,7 +21,7 @@ limitations under the License. */
* hppl pooling mode
*/
typedef
enum
{
HL_POOLING_MAX
=
0
,
HL_POOLING_MAX
=
0
,
// average includes padded values
HL_POOLING_AVERAGE
=
1
,
// average does not include padded values
...
...
@@ -324,17 +323,16 @@ extern void hl_convolution_forward_add_bias(hl_tensor_descriptor bias,
* @param[in] sizeInBytes gpu workspace size (bytes).
* @param[in] convBwdFilterAlgo backward filter algorithm.
*/
extern
void
hl_convolution_backward_filter
(
hl_tensor_descriptor
input
,
real
*
input_data
,
hl_tensor_descriptor
output
,
real
*
output_grad_data
,
hl_filter_descriptor
filter
,
real
*
filter_grad_data
,
hl_convolution_descriptor
conv
,
void
*
gpuWorkSpace
,
size_t
sizeInBytes
,
int
convBwdFilterAlgo
);
extern
void
hl_convolution_backward_filter
(
hl_tensor_descriptor
input
,
real
*
input_data
,
hl_tensor_descriptor
output
,
real
*
output_grad_data
,
hl_filter_descriptor
filter
,
real
*
filter_grad_data
,
hl_convolution_descriptor
conv
,
void
*
gpuWorkSpace
,
size_t
sizeInBytes
,
int
convBwdFilterAlgo
);
/**
* @brief convolution backward data(calculate input image grad data).
...
...
@@ -350,17 +348,16 @@ extern void hl_convolution_backward_filter(
* @param[in] sizeInBytes gpu workspace size (bytes).
* @param[in] convBwdDataAlgo backward data algorithm.
*/
extern
void
hl_convolution_backward_data
(
hl_tensor_descriptor
input
,
real
*
input_data_grad
,
hl_tensor_descriptor
output
,
real
*
output_grad_data
,
hl_filter_descriptor
filter
,
real
*
filter_data
,
hl_convolution_descriptor
conv
,
void
*
gpuWorkSpace
,
size_t
sizeInBytes
,
int
convBwdDataAlgo
);
extern
void
hl_convolution_backward_data
(
hl_tensor_descriptor
input
,
real
*
input_data_grad
,
hl_tensor_descriptor
output
,
real
*
output_grad_data
,
hl_filter_descriptor
filter
,
real
*
filter_data
,
hl_convolution_descriptor
conv
,
void
*
gpuWorkSpace
,
size_t
sizeInBytes
,
int
convBwdDataAlgo
);
/**
* @brief convolution backward bias(calculate bias grad data).
...
...
@@ -383,8 +380,8 @@ extern void hl_convolution_backward_bias(hl_tensor_descriptor bias,
* @param[in] height matrix height.
* @param[in] width matrix width.
*/
extern
void
hl_softmax_forward
(
real
*
input
,
real
*
output
,
extern
void
hl_softmax_forward
(
real
*
input
,
real
*
output
,
int
height
,
int
width
);
...
...
@@ -396,8 +393,8 @@ extern void hl_softmax_forward(real *input,
* @param[in] height matrix height.
* @param[in] width matrix width.
*/
extern
void
hl_softmax_backward
(
real
*
output_value
,
real
*
output_grad
,
extern
void
hl_softmax_backward
(
real
*
output_value
,
real
*
output_grad
,
int
height
,
int
width
);
...
...
@@ -426,18 +423,18 @@ extern void hl_softmax_backward(real *output_value,
*
*/
extern
void
hl_batch_norm_forward_training
(
hl_tensor_descriptor
inputDesc
,
real
*
input
,
real
*
input
,
hl_tensor_descriptor
outputDesc
,
real
*
output
,
real
*
output
,
hl_tensor_descriptor
bnParamDesc
,
real
*
scale
,
real
*
bias
,
real
*
scale
,
real
*
bias
,
double
factor
,
real
*
runningMean
,
real
*
runningInvVar
,
real
*
runningMean
,
real
*
runningInvVar
,
double
epsilon
,
real
*
savedMean
,
real
*
savedVar
);
real
*
savedMean
,
real
*
savedVar
);
/**
* @brief cudnn batch norm forward.
...
...
@@ -463,14 +460,14 @@ extern void hl_batch_norm_forward_training(hl_tensor_descriptor inputDesc,
*
*/
extern
void
hl_batch_norm_forward_inference
(
hl_tensor_descriptor
inputDesc
,
real
*
input
,
real
*
input
,
hl_tensor_descriptor
outputDesc
,
real
*
output
,
real
*
output
,
hl_tensor_descriptor
bnParamDesc
,
real
*
scale
,
real
*
bias
,
real
*
estimatedMean
,
real
*
estimatedVar
,
real
*
scale
,
real
*
bias
,
real
*
estimatedMean
,
real
*
estimatedVar
,
double
epsilon
);
/**
...
...
@@ -483,7 +480,8 @@ extern void hl_batch_norm_forward_inference(hl_tensor_descriptor inputDesc,
* @param[in] inGradDesc input tensor descriptor desc.
* @param[in] inGrad input data.
* @param[in] dBnParamDesc tensor descriptor desc.
* bnScale, bnBias, running mean/var, save_mean/var.
* bnScale, bnBias, running mean/var,
* save_mean/var.
* @param[in] scale batch normalization scale parameter (in original
* paper scale is referred to as gamma).
* @param[in] scaleGrad batch normalization scale parameter (in original
...
...
@@ -497,17 +495,17 @@ extern void hl_batch_norm_forward_inference(hl_tensor_descriptor inputDesc,
*
*/
extern
void
hl_batch_norm_backward
(
hl_tensor_descriptor
inputDesc
,
real
*
input
,
real
*
input
,
hl_tensor_descriptor
outGradDesc
,
real
*
outGrad
,
real
*
outGrad
,
hl_tensor_descriptor
inGradDesc
,
real
*
inGrad
,
real
*
inGrad
,
hl_tensor_descriptor
dBnParamDesc
,
real
*
scale
,
real
*
scaleGrad
,
real
*
biasGrad
,
real
*
scale
,
real
*
scaleGrad
,
real
*
biasGrad
,
double
epsilon
,
real
*
savedMean
,
real
*
savedInvVar
);
real
*
savedMean
,
real
*
savedInvVar
);
#endif // HL_CUDA_CUDNN_H_
paddle/cuda/include/hl_dso_loader.h
浏览文件 @
80c68d38
...
...
@@ -12,7 +12,6 @@ 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. */
#ifndef HL_DSO_LOADER_H_
#define HL_DSO_LOADER_H_
...
...
paddle/cuda/include/hl_functions.h
浏览文件 @
80c68d38
...
...
@@ -12,7 +12,6 @@ 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. */
#ifndef HL_FUNCTIONS_H_
#define HL_FUNCTIONS_H_
...
...
@@ -21,30 +20,30 @@ limitations under the License. */
/**
* sigmoid threshold maximum
*/
#define
SIGMOID_THRESHOLD_MIN
-40.0
#define
SIGMOID_THRESHOLD_MIN
-40.0
/**
* sigmoid threshold minimum
*/
#define
SIGMOID_THRESHOLD_MAX
13.0
#define
SIGMOID_THRESHOLD_MAX
13.0
#ifndef __NVCC__
namespace
hppl
{
/*
* forward activation
*/
real
relu
(
const
real
a
);
real
sigmoid
(
const
real
a
);
real
tanh
(
const
real
a
);
real
linear
(
const
real
a
);
/*
* backward activation
*/
real
relu
(
const
real
a
,
const
real
b
);
real
sigmoid
(
const
real
a
,
const
real
b
);
real
tanh
(
const
real
a
,
const
real
b
);
real
linear
(
const
real
a
,
const
real
b
);
/*
* forward activation
*/
real
relu
(
const
real
a
);
real
sigmoid
(
const
real
a
);
real
tanh
(
const
real
a
);
real
linear
(
const
real
a
);
/*
* backward activation
*/
real
relu
(
const
real
a
,
const
real
b
);
real
sigmoid
(
const
real
a
,
const
real
b
);
real
tanh
(
const
real
a
,
const
real
b
);
real
linear
(
const
real
a
,
const
real
b
);
}
// namespace hppl
#ifdef __AVX__
...
...
paddle/cuda/include/hl_gpu.h
浏览文件 @
80c68d38
...
...
@@ -12,7 +12,6 @@ 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. */
#ifndef HL_GPU_H_
#define HL_GPU_H_
...
...
paddle/cuda/include/hl_lstm.h
浏览文件 @
80c68d38
...
...
@@ -12,7 +12,6 @@ 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. */
#ifndef HL_LSTM_H_
#define HL_LSTM_H_
...
...
paddle/cuda/include/hl_matrix.h
浏览文件 @
80c68d38
...
...
@@ -12,7 +12,6 @@ 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. */
#ifndef HL_MATRIX_H_
#define HL_MATRIX_H_
...
...
@@ -30,13 +29,8 @@ limitations under the License. */
* @param[in] beta scalar used for addition.
*
*/
extern
void
hl_matrix_add
(
real
*
A_d
,
real
*
B_d
,
real
*
C_d
,
int
dimM
,
int
dimN
,
real
alpha
,
real
beta
);
extern
void
hl_matrix_add
(
real
*
A_d
,
real
*
B_d
,
real
*
C_d
,
int
dimM
,
int
dimN
,
real
alpha
,
real
beta
);
/**
* @brief Matrix Softmax.
*
...
...
@@ -46,7 +40,7 @@ extern void hl_matrix_add(real* A_d,
* @param[in] dimN matrix width.
*
*/
extern
void
hl_matrix_softmax
(
real
*
A_d
,
real
*
C_d
,
int
dimM
,
int
dimN
);
extern
void
hl_matrix_softmax
(
real
*
A_d
,
real
*
C_d
,
int
dimM
,
int
dimN
);
/**
* @brief Matrix softmax derivative.
...
...
@@ -58,11 +52,8 @@ extern void hl_matrix_softmax(real *A_d, real *C_d, int dimM, int dimN);
* @param[in] dimN matrix width.
*
*/
extern
void
hl_matrix_softmax_derivative
(
real
*
grad_d
,
real
*
output_d
,
real
*
sftmaxSum_d
,
int
dimM
,
int
dimN
);
extern
void
hl_matrix_softmax_derivative
(
real
*
grad_d
,
real
*
output_d
,
real
*
sftmaxSum_d
,
int
dimM
,
int
dimN
);
/**
* @brief Sequence softmax.
...
...
@@ -73,8 +64,8 @@ extern void hl_matrix_softmax_derivative(real* grad_d,
* @param[in] numSequence sequence number.
*
*/
extern
void
hl_sequence_softmax_forward
(
real
*
A_d
,
real
*
C_d
,
extern
void
hl_sequence_softmax_forward
(
real
*
A_d
,
real
*
C_d
,
const
int
*
index
,
int
numSequence
);
...
...
@@ -88,11 +79,8 @@ extern void hl_sequence_softmax_forward(real *A_d,
* @param[in] dimN matrix width.
*
*/
extern
void
hl_matrix_classification_error
(
real
*
A_d
,
int
*
B_d
,
real
*
C_d
,
int
dimM
,
int
dimN
);
extern
void
hl_matrix_classification_error
(
real
*
A_d
,
int
*
B_d
,
real
*
C_d
,
int
dimM
,
int
dimN
);
/**
* @brief Matrix cross entropy.
...
...
@@ -104,11 +92,8 @@ extern void hl_matrix_classification_error(real* A_d,
* @param[in] dimN matrix width.
*
*/
extern
void
hl_matrix_cross_entropy
(
real
*
A_d
,
real
*
C_d
,
int
*
label_d
,
int
dimM
,
int
dimN
);
extern
void
hl_matrix_cross_entropy
(
real
*
A_d
,
real
*
C_d
,
int
*
label_d
,
int
dimM
,
int
dimN
);
/**
* @brief Matrix cross entropy back propagation.
...
...
@@ -120,11 +105,8 @@ extern void hl_matrix_cross_entropy(real* A_d,
* @param[in] dimN matrix width.
*
*/
extern
void
hl_matrix_cross_entropy_bp
(
real
*
grad_d
,
real
*
output_d
,
int
*
label_d
,
int
dimM
,
int
dimN
);
extern
void
hl_matrix_cross_entropy_bp
(
real
*
grad_d
,
real
*
output_d
,
int
*
label_d
,
int
dimM
,
int
dimN
);
/**
* @brief Matrix multi-binary label cross entropy
...
...
@@ -135,11 +117,8 @@ extern void hl_matrix_cross_entropy_bp(real* grad_d,
* @param[in] dimM matrix height.
* @param[in] dimN matrix width.
*/
extern
void
hl_matrix_multi_binary_cross_entropy
(
real
*
output
,
real
*
entropy
,
hl_sparse_matrix_s
mat
,
int
dimM
,
int
dimN
);
extern
void
hl_matrix_multi_binary_cross_entropy
(
real
*
output
,
real
*
entropy
,
hl_sparse_matrix_s
mat
,
int
dimM
,
int
dimN
);
/**
* @brief Matrix multi-binary label cross entropy backprop
...
...
@@ -150,11 +129,8 @@ extern void hl_matrix_multi_binary_cross_entropy(real* output,
* @param[in] dimM matrix height.
* @param[in] dimN matrix width.
*/
extern
void
hl_matrix_multi_binary_cross_entropy_bp
(
real
*
output
,
real
*
grad
,
hl_sparse_matrix_s
mat
,
int
dimM
,
int
dimN
);
extern
void
hl_matrix_multi_binary_cross_entropy_bp
(
real
*
output
,
real
*
grad
,
hl_sparse_matrix_s
mat
,
int
dimM
,
int
dimN
);
/**
* @brief Matrix zero memory.
...
...
@@ -176,12 +152,8 @@ extern void hl_matrix_zero_mem(real* data, int num);
* @param[in] partial_sum
*/
extern
void
hl_param_relu_forward
(
real
*
output
,
real
*
input
,
real
*
w
,
int
width
,
int
height
,
int
partial_sum
);
extern
void
hl_param_relu_forward
(
real
*
output
,
real
*
input
,
real
*
w
,
int
width
,
int
height
,
int
partial_sum
);
/**
* @brief parameter relu backward w
*
...
...
paddle/cuda/include/hl_sequence.h
浏览文件 @
80c68d38
...
...
@@ -12,7 +12,6 @@ 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. */
#ifndef HL_SEQUENCE_H_
#define HL_SEQUENCE_H_
...
...
@@ -32,7 +31,7 @@ limitations under the License. */
extern
void
hl_max_sequence_forward
(
real
*
input
,
const
int
*
sequence
,
real
*
output
,
int
*
index
,
int
*
index
,
int
numSequences
,
int
dim
);
...
...
@@ -46,11 +45,8 @@ extern void hl_max_sequence_forward(real* input,
* @param[in] dim input dimension.
*
*/
extern
void
hl_max_sequence_backward
(
real
*
outputGrad
,
int
*
index
,
real
*
inputGrad
,
int
numSequences
,
int
dim
);
extern
void
hl_max_sequence_backward
(
real
*
outputGrad
,
int
*
index
,
real
*
inputGrad
,
int
numSequences
,
int
dim
);
/**
* @brief Context projection forward.
...
...
@@ -63,7 +59,8 @@ extern void hl_max_sequence_backward(real* outputGrad,
* @param[in] inputDim input sequence dimension.
* @param[in] contextLength context length.
* @param[in] contextStart context start.
* @param[in] beginPad number of extra timesteps added at the beginning.
* @param[in] beginPad number of extra timesteps added at the
* beginning.
* @param[in] isPadding trainable padding.
*
*/
...
...
@@ -109,7 +106,8 @@ extern void hl_context_projection_backward_data(real* outputGrad,
* @param[in] totalPad number of extra timesteps.
* @param[in] contextLength context length.
* @param[in] contextStart context start.
* @param[in] beginPad number of extra timesteps added at the beginning.
* @param[in] beginPad number of extra timesteps added at the
* beginning.
*
*/
extern
void
hl_context_projection_backward_weight
(
real
*
outputGrad
,
...
...
@@ -141,9 +139,9 @@ extern void hl_context_projection_backward_weight(real* outputGrad,
* @param[in] seq2batch copy direction.
*
*/
extern
void
hl_sequence2batch_copy
(
real
*
batch
,
real
*
sequence
,
const
int
*
batchIndex
,
extern
void
hl_sequence2batch_copy
(
real
*
batch
,
real
*
sequence
,
const
int
*
batchIndex
,
int
seqWidth
,
int
batchCount
,
bool
seq2batch
);
...
...
@@ -167,9 +165,9 @@ extern void hl_sequence2batch_copy(real *batch,
* @param[in] seq2batch copy direction.
*
*/
extern
void
hl_sequence2batch_add
(
real
*
batch
,
real
*
sequence
,
int
*
batchIndex
,
extern
void
hl_sequence2batch_add
(
real
*
batch
,
real
*
sequence
,
int
*
batchIndex
,
int
seqWidth
,
int
batchCount
,
bool
seq2batch
);
...
...
paddle/cuda/include/hl_sparse.h
浏览文件 @
80c68d38
...
...
@@ -12,7 +12,6 @@ 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. */
#ifndef HL_SPARSE_H_
#define HL_SPARSE_H_
...
...
@@ -31,7 +30,7 @@ limitations under the License. */
*/
extern
void
hl_malloc_sparse_matrix
(
hl_sparse_matrix_s
*
A_d
,
hl_matrix_format_t
format
,
hl_matrix_value_t
value_type
,
hl_matrix_value_t
value_type
,
int
dimM
,
int
dimN
,
int
nnz
);
...
...
@@ -60,10 +59,10 @@ extern void hl_free_sparse_matrix(hl_sparse_matrix_s A_d);
*
*/
extern
void
hl_construct_sparse_matrix
(
hl_sparse_matrix_s
*
A_d
,
void
*
dest_d
,
void
*
dest_d
,
size_t
size
,
hl_matrix_format_t
format
,
hl_matrix_value_t
value_type
,
hl_matrix_value_t
value_type
,
int
dimM
,
int
dimN
,
int
nnz
);
...
...
@@ -94,11 +93,11 @@ extern void hl_construct_sparse_matrix(hl_sparse_matrix_s *A_d,
*
*/
extern
void
hl_construct_sparse_matrix
(
hl_sparse_matrix_s
*
A_d
,
real
*
value_d
,
int
*
rows_d
,
int
*
cols_d
,
real
*
value_d
,
int
*
rows_d
,
int
*
cols_d
,
hl_matrix_format_t
format
,
hl_matrix_value_t
value_type
,
hl_matrix_value_t
value_type
,
int
dimM
,
int
dimN
,
int
nnz
);
...
...
@@ -259,10 +258,14 @@ extern void hl_matrix_csr_mul_dense(hl_sparse_matrix_s A_d,
*/
extern
void
hl_matrix_csc_mul_dense
(
hl_sparse_matrix_s
A_d
,
hl_trans_op_t
transa
,
real
*
B_d
,
hl_trans_op_t
transb
,
real
*
B_d
,
hl_trans_op_t
transb
,
real
*
C_d
,
int
dimM
,
int
dimN
,
int
dimK
,
real
alpha
,
real
beta
);
int
dimM
,
int
dimN
,
int
dimK
,
real
alpha
,
real
beta
);
/**
* @brief C_d = alpha*(op(A_d) * op(B_d)) + beta*C_d.
...
...
@@ -311,11 +314,16 @@ extern void hl_matrix_dense_mul_csc(real *A_d,
* @note transb is not support HPPL_OP_T.
*
*/
extern
void
hl_sparse_matrix_mul
(
real
*
A_d
,
hl_trans_op_t
transa
,
real
*
B_d
,
hl_trans_op_t
transb
,
extern
void
hl_sparse_matrix_mul
(
real
*
A_d
,
hl_trans_op_t
transa
,
real
*
B_d
,
hl_trans_op_t
transb
,
hl_sparse_matrix_s
C_d
,
int
dimM
,
int
dimN
,
int
dimK
,
real
alpha
,
real
beta
);
int
dimM
,
int
dimN
,
int
dimK
,
real
alpha
,
real
beta
);
/**
* @brief C_d = alpha*(op(A_d) * op(B_d)) + beta*C_d
...
...
@@ -336,12 +344,16 @@ extern void hl_sparse_matrix_mul(real* A_d, hl_trans_op_t transa,
* @note transa is not support HPPL_OP_T.
*
*/
extern
void
hl_matrix_dense_mul_csr
(
real
*
A_d
,
hl_trans_op_t
transa
,
extern
void
hl_matrix_dense_mul_csr
(
real
*
A_d
,
hl_trans_op_t
transa
,
hl_sparse_matrix_s
B_d
,
hl_trans_op_t
transb
,
real
*
C_d
,
int
dimM
,
int
dimN
,
int
dimK
,
real
alpha
,
real
beta
);
int
dimM
,
int
dimN
,
int
dimK
,
real
alpha
,
real
beta
);
/**
* @brief Memcpy csc_matrix to host.
...
...
@@ -412,7 +424,6 @@ extern void hl_memcpy_from_csr_matrix(real *csr_val,
hl_sparse_matrix_s
csr_matrix
,
hl_stream_t
stream
);
/**
* @brief A_d[j] += B_d[i,j] for i in range(height)
*
...
...
@@ -423,19 +434,13 @@ extern void hl_memcpy_from_csr_matrix(real *csr_val,
* @param[in] scale scale of B_d
*
*/
extern
void
hl_sparse_matrix_column_sum
(
real
*
A_d
,
hl_sparse_matrix_s
B_d
,
int
dimM
,
int
dimN
,
real
scale
);
extern
void
hl_sparse_matrix_column_sum
(
real
*
A_d
,
hl_sparse_matrix_s
B_d
,
int
dimM
,
int
dimN
,
real
scale
);
/**
* @brief implementation of csr sparse matrix in hl_sparse_matirx_column_sum
*/
extern
void
hl_matrix_csr_column_sum
(
real
*
A_d
,
hl_sparse_matrix_s
B_d
,
int
dimM
,
int
dimN
,
real
scale
);
extern
void
hl_matrix_csr_column_sum
(
real
*
A_d
,
hl_sparse_matrix_s
B_d
,
int
dimM
,
int
dimN
,
real
scale
);
/**
* @brief A_d[i,j] += B_d[j]
...
...
@@ -446,13 +451,13 @@ extern void hl_matrix_csr_column_sum(real* A_d,
*
*/
extern
void
hl_sparse_matrix_add_bias
(
hl_sparse_matrix_s
A_d
,
real
*
B_d
,
real
*
B_d
,
real
scale
);
/**
* @brief implementation of csr sparse matrix in hl_sparse_matrix_add_bias
*/
extern
void
hl_matrix_csr_add_bias
(
hl_sparse_matrix_s
A_d
,
real
*
B_d
,
real
*
B_d
,
real
scale
);
/**
...
...
@@ -470,7 +475,7 @@ extern void hl_matrix_csr_add_bias(hl_sparse_matrix_s A_d,
*
*/
extern
void
hl_sparse_matrix_add_dense
(
hl_sparse_matrix_s
A_d
,
real
*
B_d
,
real
*
B_d
,
int
dimM
,
int
dimN
,
real
alpha
,
...
...
@@ -479,7 +484,7 @@ extern void hl_sparse_matrix_add_dense(hl_sparse_matrix_s A_d,
* @brief implementation of csr sparse matrix in hl_sparse_matrix_add_dense
*/
extern
void
hl_matrix_csr_add_dense
(
hl_sparse_matrix_s
A_d
,
real
*
B_d
,
real
*
B_d
,
int
dimM
,
int
dimN
,
real
alpha
,
...
...
@@ -493,7 +498,7 @@ extern void hl_matrix_csr_add_dense(hl_sparse_matrix_s A_d,
* @return return rows pointer, which is gpu address
*
*/
extern
int
*
hl_sparse_matrix_get_rows
(
hl_sparse_matrix_s
sMat
);
extern
int
*
hl_sparse_matrix_get_rows
(
hl_sparse_matrix_s
sMat
);
/**
* @brief get cols pionter of GpuSparseMatrix
...
...
@@ -503,7 +508,7 @@ extern int* hl_sparse_matrix_get_rows(hl_sparse_matrix_s sMat);
* @return return cols pointer, which is gpu address
*
*/
extern
int
*
hl_sparse_matrix_get_cols
(
hl_sparse_matrix_s
sMat
);
extern
int
*
hl_sparse_matrix_get_cols
(
hl_sparse_matrix_s
sMat
);
/**
* @brief get value pionter of GpuSparseMatrix
...
...
@@ -513,7 +518,6 @@ extern int* hl_sparse_matrix_get_cols(hl_sparse_matrix_s sMat);
* @return return value pointer, which is gpu address
*
*/
extern
real
*
hl_sparse_matrix_get_value
(
hl_sparse_matrix_s
sMat
);
extern
real
*
hl_sparse_matrix_get_value
(
hl_sparse_matrix_s
sMat
);
#endif
/* HL_SPARSE_H_ */
paddle/cuda/include/hl_table_apply.h
浏览文件 @
80c68d38
...
...
@@ -12,7 +12,6 @@ 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. */
#ifndef HL_TABLE_APPLY_H_
#define HL_TABLE_APPLY_H_
...
...
@@ -31,8 +30,10 @@ limitations under the License. */
* @param[in] dim width of table.
*
*/
extern
void
hl_matrix_select_rows
(
real
*
output
,
int
ldo
,
real
*
table
,
int
ldt
,
extern
void
hl_matrix_select_rows
(
real
*
output
,
int
ldo
,
real
*
table
,
int
ldt
,
int
*
ids
,
int
numSamples
,
int
tableSize
,
...
...
@@ -53,8 +54,10 @@ extern void hl_matrix_select_rows(real* output, int ldo,
* @param[in] dim width of table.
*
*/
extern
void
hl_matrix_add_to_rows
(
real
*
table
,
int
ldt
,
real
*
input
,
int
ldi
,
extern
void
hl_matrix_add_to_rows
(
real
*
table
,
int
ldt
,
real
*
input
,
int
ldi
,
int
*
ids
,
int
numSamples
,
int
tableSize
,
...
...
@@ -72,8 +75,7 @@ extern void hl_matrix_add_to_rows(real* table, int ldt,
*
*/
template
<
class
T
>
extern
void
hl_vector_select_from
(
T
*
dst
,
int
sized
,
const
T
*
src
,
int
sizes
,
const
int
*
ids
,
int
sizei
);
extern
void
hl_vector_select_from
(
T
*
dst
,
int
sized
,
const
T
*
src
,
int
sizes
,
const
int
*
ids
,
int
sizei
);
#endif
/* HL_TABLE_APPLY_H_ */
#endif
/* HL_TABLE_APPLY_H_ */
paddle/cuda/include/hl_time.h
浏览文件 @
80c68d38
...
...
@@ -12,7 +12,6 @@ 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. */
#ifndef HL_TIME_H_
#define HL_TIME_H_
...
...
paddle/cuda/include/hl_top_k.h
浏览文件 @
80c68d38
...
...
@@ -12,7 +12,6 @@ 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. */
#ifndef HL_TOP_K_H_
#define HL_TOP_K_H_
...
...
@@ -31,9 +30,11 @@ limitations under the License. */
* @param[in] numSamples height of input value.
*
*/
extern
void
hl_matrix_top_k
(
real
*
topVal
,
int
ldv
,
int
*
topIds
,
real
*
src
,
int
lds
,
extern
void
hl_matrix_top_k
(
real
*
topVal
,
int
ldv
,
int
*
topIds
,
real
*
src
,
int
lds
,
int
dim
,
int
beamSize
,
int
numSamples
);
...
...
@@ -50,8 +51,9 @@ extern void hl_matrix_top_k(real* topVal, int ldv,
*
* @note Only support HL_SPARSE_CSR format.
*/
extern
void
hl_sparse_matrix_top_k
(
real
*
topVal
,
int
ldv
,
int
*
topIds
,
extern
void
hl_sparse_matrix_top_k
(
real
*
topVal
,
int
ldv
,
int
*
topIds
,
hl_sparse_matrix_s
src
,
int
beamSize
,
int
numSamples
);
...
...
paddle/cuda/include/stub/hl_aggregate_stub.h
浏览文件 @
80c68d38
...
...
@@ -12,29 +12,22 @@ 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. */
#ifndef HL_AGGREGATE_STUB_H_
#define HL_AGGREGATE_STUB_H_
#include "hl_aggregate.h"
inline
void
hl_matrix_row_sum
(
real
*
A_d
,
real
*
C_d
,
int
dimM
,
int
dimN
)
{}
inline
void
hl_matrix_row_sum
(
real
*
A_d
,
real
*
C_d
,
int
dimM
,
int
dimN
)
{}
inline
void
hl_matrix_row_max
(
real
*
A_d
,
real
*
C_d
,
int
dimM
,
int
dimN
)
{}
inline
void
hl_matrix_row_max
(
real
*
A_d
,
real
*
C_d
,
int
dimM
,
int
dimN
)
{}
inline
void
hl_matrix_row_min
(
real
*
A_d
,
real
*
C_d
,
int
dimM
,
int
dimN
)
{}
inline
void
hl_matrix_row_min
(
real
*
A_d
,
real
*
C_d
,
int
dimM
,
int
dimN
)
{}
inline
void
hl_matrix_column_sum
(
real
*
A_d
,
real
*
C_d
,
int
dimM
,
int
dimN
)
{}
inline
void
hl_matrix_column_sum
(
real
*
A_d
,
real
*
C_d
,
int
dimM
,
int
dimN
)
{}
inline
void
hl_matrix_column_max
(
real
*
A_d
,
real
*
C_d
,
int
dimM
,
int
dimN
)
{}
inline
void
hl_matrix_column_max
(
real
*
A_d
,
real
*
C_d
,
int
dimM
,
int
dimN
)
{}
inline
void
hl_matrix_column_min
(
real
*
A_d
,
real
*
C_d
,
int
dimM
,
int
dimN
)
{}
inline
void
hl_matrix_column_min
(
real
*
A_d
,
real
*
C_d
,
int
dimM
,
int
dimN
)
{}
inline
void
hl_vector_sum
(
real
*
A_d
,
real
*
C_h
,
int
dimM
)
{}
...
...
paddle/cuda/include/stub/hl_cnn_stub.h
浏览文件 @
80c68d38
...
...
@@ -12,84 +12,134 @@ 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. */
#ifndef HL_CNN_STUB_H_
#define HL_CNN_STUB_H_
#include "hl_cnn.h"
inline
void
hl_shrink_col2feature
(
const
real
*
dataCol
,
size_t
channels
,
size_t
height
,
size_t
width
,
size_t
blockH
,
size_t
blockW
,
size_t
strideH
,
size_t
strideW
,
size_t
paddingH
,
size_t
paddingW
,
size_t
outputH
,
size_t
outputW
,
real
*
dataIm
,
real
alpha
,
real
beta
)
{}
inline
void
hl_expand_feature2col
(
const
real
*
dataIm
,
size_t
channels
,
size_t
height
,
size_t
width
,
size_t
blockH
,
size_t
blockW
,
size_t
strideH
,
size_t
strideW
,
size_t
paddingH
,
size_t
paddingW
,
size_t
outputH
,
size_t
outputW
,
real
*
dataCol
)
{}
inline
void
hl_maxpool_forward
(
const
int
frameCnt
,
const
real
*
inputData
,
const
int
channels
,
const
int
height
,
const
int
width
,
const
int
pooledH
,
const
int
pooledW
,
const
int
sizeX
,
const
int
sizeY
,
const
int
strideH
,
const
int
strideW
,
const
int
paddingH
,
const
int
paddingW
,
real
*
tgtData
,
const
int
tgtStride
)
{}
inline
void
hl_maxpool_backward
(
const
int
frameCnt
,
const
real
*
inputData
,
const
real
*
outData
,
const
real
*
outGrad
,
const
int
channels
,
const
int
height
,
const
int
width
,
const
int
pooledH
,
const
int
pooledW
,
const
int
sizeX
,
const
int
sizeY
,
const
int
strideH
,
const
int
strideW
,
const
int
paddingH
,
const
int
paddingW
,
real
scaleA
,
real
scaleB
,
real
*
targetGrad
,
const
int
outStride
)
{}
inline
void
hl_avgpool_forward
(
const
int
frameCnt
,
const
real
*
inputData
,
const
int
channels
,
const
int
height
,
const
int
width
,
const
int
pooledH
,
const
int
pooledW
,
const
int
sizeX
,
const
int
sizeY
,
const
int
strideH
,
const
int
strideW
,
const
int
paddingH
,
const
int
paddingW
,
real
*
tgtData
,
const
int
tgtStride
)
{}
inline
void
hl_avgpool_backward
(
const
int
frameCnt
,
const
real
*
outGrad
,
const
int
channels
,
const
int
height
,
const
int
width
,
const
int
pooledH
,
const
int
pooledW
,
const
int
sizeX
,
const
int
sizeY
,
const
int
strideH
,
const
int
strideW
,
int
paddingH
,
int
paddingW
,
real
scaleA
,
real
scaleB
,
real
*
backGrad
,
const
int
outStride
)
{}
inline
void
hl_CMRNorm_forward
(
size_t
frameCnt
,
const
real
*
in
,
real
*
scale
,
real
*
out
,
size_t
channels
,
size_t
height
,
size_t
width
,
size_t
sizeX
,
real
alpha
,
real
beta
)
{}
inline
void
hl_CMRNorm_backward
(
size_t
frameCnt
,
const
real
*
inV
,
const
real
*
scale
,
const
real
*
outV
,
const
real
*
outDiff
,
real
*
inDiff
,
size_t
channels
,
size_t
height
,
size_t
width
,
size_t
sizeX
,
real
alpha
,
real
beta
)
{}
inline
void
hl_shrink_col2feature
(
const
real
*
dataCol
,
size_t
channels
,
size_t
height
,
size_t
width
,
size_t
blockH
,
size_t
blockW
,
size_t
strideH
,
size_t
strideW
,
size_t
paddingH
,
size_t
paddingW
,
size_t
outputH
,
size_t
outputW
,
real
*
dataIm
,
real
alpha
,
real
beta
)
{}
inline
void
hl_expand_feature2col
(
const
real
*
dataIm
,
size_t
channels
,
size_t
height
,
size_t
width
,
size_t
blockH
,
size_t
blockW
,
size_t
strideH
,
size_t
strideW
,
size_t
paddingH
,
size_t
paddingW
,
size_t
outputH
,
size_t
outputW
,
real
*
dataCol
)
{}
inline
void
hl_maxpool_forward
(
const
int
frameCnt
,
const
real
*
inputData
,
const
int
channels
,
const
int
height
,
const
int
width
,
const
int
pooledH
,
const
int
pooledW
,
const
int
sizeX
,
const
int
sizeY
,
const
int
strideH
,
const
int
strideW
,
const
int
paddingH
,
const
int
paddingW
,
real
*
tgtData
,
const
int
tgtStride
)
{}
inline
void
hl_maxpool_backward
(
const
int
frameCnt
,
const
real
*
inputData
,
const
real
*
outData
,
const
real
*
outGrad
,
const
int
channels
,
const
int
height
,
const
int
width
,
const
int
pooledH
,
const
int
pooledW
,
const
int
sizeX
,
const
int
sizeY
,
const
int
strideH
,
const
int
strideW
,
const
int
paddingH
,
const
int
paddingW
,
real
scaleA
,
real
scaleB
,
real
*
targetGrad
,
const
int
outStride
)
{}
inline
void
hl_avgpool_forward
(
const
int
frameCnt
,
const
real
*
inputData
,
const
int
channels
,
const
int
height
,
const
int
width
,
const
int
pooledH
,
const
int
pooledW
,
const
int
sizeX
,
const
int
sizeY
,
const
int
strideH
,
const
int
strideW
,
const
int
paddingH
,
const
int
paddingW
,
real
*
tgtData
,
const
int
tgtStride
)
{}
inline
void
hl_avgpool_backward
(
const
int
frameCnt
,
const
real
*
outGrad
,
const
int
channels
,
const
int
height
,
const
int
width
,
const
int
pooledH
,
const
int
pooledW
,
const
int
sizeX
,
const
int
sizeY
,
const
int
strideH
,
const
int
strideW
,
int
paddingH
,
int
paddingW
,
real
scaleA
,
real
scaleB
,
real
*
backGrad
,
const
int
outStride
)
{}
inline
void
hl_CMRNorm_forward
(
size_t
frameCnt
,
const
real
*
in
,
real
*
scale
,
real
*
out
,
size_t
channels
,
size_t
height
,
size_t
width
,
size_t
sizeX
,
real
alpha
,
real
beta
)
{}
inline
void
hl_CMRNorm_backward
(
size_t
frameCnt
,
const
real
*
inV
,
const
real
*
scale
,
const
real
*
outV
,
const
real
*
outDiff
,
real
*
inDiff
,
size_t
channels
,
size_t
height
,
size_t
width
,
size_t
sizeX
,
real
alpha
,
real
beta
)
{}
inline
void
hl_bilinear_forward
(
const
real
*
inData
,
const
size_t
inImgH
,
...
...
@@ -106,25 +156,33 @@ inline void hl_bilinear_forward(const real* inData,
const
real
ratioW
)
{}
inline
void
hl_bilinear_backward
(
real
*
inGrad
,
const
size_t
inImgH
,
const
size_t
inImgW
,
const
size_t
inputH
,
const
size_t
inputW
,
const
real
*
outGrad
,
const
size_t
outImgH
,
const
size_t
outImgW
,
const
size_t
outputH
,
const
size_t
outputW
,
const
size_t
numChannels
,
const
real
ratioH
,
const
real
ratioW
)
{}
inline
void
hl_maxout_forward
(
const
real
*
inData
,
real
*
outData
,
int
*
idData
,
size_t
batchSize
,
size_t
size
,
size_t
featLen
,
size_t
group
)
{}
inline
void
hl_maxout_backward
(
real
*
inGrad
,
const
real
*
outGrad
,
const
int
*
idData
,
size_t
batchSize
,
size_t
size
,
size_t
featLen
,
size_t
group
)
{}
const
size_t
inImgH
,
const
size_t
inImgW
,
const
size_t
inputH
,
const
size_t
inputW
,
const
real
*
outGrad
,
const
size_t
outImgH
,
const
size_t
outImgW
,
const
size_t
outputH
,
const
size_t
outputW
,
const
size_t
numChannels
,
const
real
ratioH
,
const
real
ratioW
)
{}
inline
void
hl_maxout_forward
(
const
real
*
inData
,
real
*
outData
,
int
*
idData
,
size_t
batchSize
,
size_t
size
,
size_t
featLen
,
size_t
group
)
{}
inline
void
hl_maxout_backward
(
real
*
inGrad
,
const
real
*
outGrad
,
const
int
*
idData
,
size_t
batchSize
,
size_t
size
,
size_t
featLen
,
size_t
group
)
{}
#endif // HL_CNN_STUB_H_
paddle/cuda/include/stub/hl_cuda_cublas_stub.h
浏览文件 @
80c68d38
...
...
@@ -12,41 +12,42 @@ 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. */
#ifndef HL_CUDA_CUBLAS_STUB_H_
#define HL_CUDA_CUBLAS_STUB_H_
#include "hl_cuda_cublas.h"
inline
void
hl_matrix_transpose
(
real
*
A_d
,
real
*
C_d
,
int
dimM
,
int
dimN
,
int
lda
,
int
ldc
)
{}
inline
void
hl_matrix_transpose
(
real
*
A_d
,
real
*
C_d
,
int
dimM
,
int
dimN
)
{}
inline
void
hl_matrix_inverse
(
real
*
A_d
,
real
*
C_d
,
int
dimN
,
int
lda
,
int
ldc
)
{}
inline
void
hl_matrix_mul
(
real
*
A_d
,
hl_trans_op_t
transa
,
real
*
B_d
,
hl_trans_op_t
transb
,
real
*
C_d
,
int
dimM
,
int
dimN
,
int
dimK
,
real
alpha
,
real
beta
,
int
lda
,
int
ldb
,
int
ldc
)
{}
inline
void
hl_matrix_transpose
(
real
*
A_d
,
real
*
C_d
,
int
dimM
,
int
dimN
,
int
lda
,
int
ldc
)
{}
inline
void
hl_matrix_transpose
(
real
*
A_d
,
real
*
C_d
,
int
dimM
,
int
dimN
)
{}
inline
void
hl_matrix_mul
(
real
*
A_d
,
hl_trans_op_t
transa
,
real
*
B_d
,
hl_trans_op_t
transb
,
inline
void
hl_matrix_inverse
(
real
*
A_d
,
real
*
C_d
,
int
dimN
,
int
lda
,
int
ldc
)
{}
inline
void
hl_matrix_mul
(
real
*
A_d
,
hl_trans_op_t
transa
,
real
*
B_d
,
hl_trans_op_t
transb
,
real
*
C_d
,
int
dimM
,
int
dimN
,
int
dimK
,
real
alpha
,
real
beta
,
int
lda
,
int
ldb
,
int
ldc
)
{}
inline
void
hl_matrix_mul
(
real
*
A_d
,
hl_trans_op_t
transa
,
real
*
B_d
,
hl_trans_op_t
transb
,
real
*
C_d
,
int
dimM
,
int
dimN
,
int
dimK
,
real
alpha
,
real
beta
)
{}
int
dimM
,
int
dimN
,
int
dimK
,
real
alpha
,
real
beta
)
{}
#endif // HL_CUDA_CUBLAS_STUB_H_
paddle/cuda/include/stub/hl_cuda_cudnn_stub.h
浏览文件 @
80c68d38
...
...
@@ -12,15 +12,12 @@ 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. */
#ifndef HL_CUDA_CUDNN_STUB_H_
#define HL_CUDA_CUDNN_STUB_H_
#include "hl_cuda_cudnn.h"
inline
int
hl_get_cudnn_lib_version
()
{
return
0
;
}
inline
int
hl_get_cudnn_lib_version
()
{
return
0
;
}
inline
void
hl_create_tensor_descriptor
(
hl_tensor_descriptor
*
image_desc
)
{}
...
...
@@ -68,41 +65,41 @@ inline void hl_pooling_backward(hl_tensor_descriptor input,
hl_pooling_descriptor
pooling
)
{}
inline
void
hl_create_filter_descriptor
(
hl_filter_descriptor
*
filter
,
int
input_feature_maps
,
int
output_feature_maps
,
int
height
,
int
width
)
{}
int
input_feature_maps
,
int
output_feature_maps
,
int
height
,
int
width
)
{}
inline
void
hl_destroy_filter_descriptor
(
hl_filter_descriptor
filter
)
{}
inline
void
hl_create_convolution_descriptor
(
hl_convolution_descriptor
*
conv
,
hl_tensor_descriptor
image
,
hl_filter_descriptor
filter
,
int
padding_height
,
int
padding_width
,
int
stride_height
,
int
stride_width
)
{}
hl_tensor_descriptor
image
,
hl_filter_descriptor
filter
,
int
padding_height
,
int
padding_width
,
int
stride_height
,
int
stride_width
)
{}
inline
void
hl_reset_convolution_descriptor
(
hl_convolution_descriptor
conv
,
hl_tensor_descriptor
image
,
hl_filter_descriptor
filter
,
int
padding_height
,
int
padding_width
,
int
stride_height
,
int
stride_width
)
{}
hl_tensor_descriptor
image
,
hl_filter_descriptor
filter
,
int
padding_height
,
int
padding_width
,
int
stride_height
,
int
stride_width
)
{}
inline
void
hl_destroy_convolution_descriptor
(
hl_convolution_descriptor
conv
)
{}
inline
void
hl_conv_workspace
(
hl_tensor_descriptor
input
,
hl_tensor_descriptor
output
,
hl_filter_descriptor
filter
,
hl_convolution_descriptor
conv
,
int
*
convFwdAlgo
,
size_t
*
fwdLimitBytes
,
int
*
convBwdDataAlgo
,
size_t
*
bwdDataLimitBytes
,
int
*
convBwdFilterAlgo
,
size_t
*
bwdFilterLimitBytes
)
{}
hl_tensor_descriptor
output
,
hl_filter_descriptor
filter
,
hl_convolution_descriptor
conv
,
int
*
convFwdAlgo
,
size_t
*
fwdLimitBytes
,
int
*
convBwdDataAlgo
,
size_t
*
bwdDataLimitBytes
,
int
*
convBwdFilterAlgo
,
size_t
*
bwdFilterLimitBytes
)
{}
inline
void
hl_convolution_forward
(
hl_tensor_descriptor
input
,
real
*
input_data
,
...
...
@@ -116,86 +113,84 @@ inline void hl_convolution_forward(hl_tensor_descriptor input,
int
convFwdAlgo
)
{}
inline
void
hl_convolution_forward_add_bias
(
hl_tensor_descriptor
bias
,
real
*
bias_data
,
hl_tensor_descriptor
output
,
real
*
output_data
)
{}
inline
void
hl_convolution_backward_filter
(
hl_tensor_descriptor
input
,
real
*
input_data
,
hl_tensor_descriptor
output
,
real
*
output_grad_data
,
hl_filter_descriptor
filter
,
real
*
filter_grad_data
,
hl_convolution_descriptor
conv
,
void
*
gpuWorkSpace
,
size_t
sizeInBytes
,
int
convBwdFilterAlgo
)
{}
inline
void
hl_convolution_backward_data
(
hl_tensor_descriptor
input
,
real
*
input_data_grad
,
hl_tensor_descriptor
output
,
real
*
output_grad_data
,
hl_filter_descriptor
filter
,
real
*
filter_data
,
hl_convolution_descriptor
conv
,
void
*
gpuWorkSpace
,
size_t
sizeInBytes
,
int
convBwdDataAlgo
)
{}
real
*
bias_data
,
hl_tensor_descriptor
output
,
real
*
output_data
)
{}
inline
void
hl_convolution_backward_filter
(
hl_tensor_descriptor
input
,
real
*
input_data
,
hl_tensor_descriptor
output
,
real
*
output_grad_data
,
hl_filter_descriptor
filter
,
real
*
filter_grad_data
,
hl_convolution_descriptor
conv
,
void
*
gpuWorkSpace
,
size_t
sizeInBytes
,
int
convBwdFilterAlgo
)
{}
inline
void
hl_convolution_backward_data
(
hl_tensor_descriptor
input
,
real
*
input_data_grad
,
hl_tensor_descriptor
output
,
real
*
output_grad_data
,
hl_filter_descriptor
filter
,
real
*
filter_data
,
hl_convolution_descriptor
conv
,
void
*
gpuWorkSpace
,
size_t
sizeInBytes
,
int
convBwdDataAlgo
)
{}
inline
void
hl_convolution_backward_bias
(
hl_tensor_descriptor
bias
,
real
*
bias_grad_data
,
hl_tensor_descriptor
output
,
real
*
output_grad_data
)
{}
real
*
bias_grad_data
,
hl_tensor_descriptor
output
,
real
*
output_grad_data
)
{}
inline
void
hl_softmax_forward
(
real
*
input
,
real
*
output
,
int
height
,
int
width
)
{}
inline
void
hl_softmax_backward
(
real
*
output_value
,
real
*
output_grad
,
inline
void
hl_softmax_forward
(
real
*
input
,
real
*
output
,
int
height
,
int
width
)
{}
inline
void
hl_softmax_backward
(
real
*
output_value
,
real
*
output_grad
,
int
height
,
int
width
)
{}
inline
void
hl_batch_norm_forward_training
(
hl_tensor_descriptor
inputDesc
,
real
*
input
,
real
*
input
,
hl_tensor_descriptor
outputDesc
,
real
*
output
,
real
*
output
,
hl_tensor_descriptor
bnParamDesc
,
real
*
scale
,
real
*
bias
,
real
*
scale
,
real
*
bias
,
double
factor
,
real
*
runningMean
,
real
*
runningInvVar
,
real
*
runningMean
,
real
*
runningInvVar
,
double
epsilon
,
real
*
savedMean
,
real
*
savedVar
)
{}
real
*
savedMean
,
real
*
savedVar
)
{}
inline
void
hl_batch_norm_forward_inference
(
hl_tensor_descriptor
inputDesc
,
real
*
input
,
real
*
input
,
hl_tensor_descriptor
outputDesc
,
real
*
output
,
real
*
output
,
hl_tensor_descriptor
bnParamDesc
,
real
*
scale
,
real
*
bias
,
real
*
estimatedMean
,
real
*
estimatedVar
,
real
*
scale
,
real
*
bias
,
real
*
estimatedMean
,
real
*
estimatedVar
,
double
epsilon
)
{}
inline
void
hl_batch_norm_backward
(
hl_tensor_descriptor
inputDesc
,
real
*
input
,
real
*
input
,
hl_tensor_descriptor
outGradDesc
,
real
*
outGrad
,
real
*
outGrad
,
hl_tensor_descriptor
inGradDesc
,
real
*
inGrad
,
real
*
inGrad
,
hl_tensor_descriptor
dBnParamDesc
,
real
*
scale
,
real
*
scaleGrad
,
real
*
biasGrad
,
real
*
scale
,
real
*
scaleGrad
,
real
*
biasGrad
,
double
epsilon
,
real
*
savedMean
,
real
*
savedInvVar
)
{}
real
*
savedMean
,
real
*
savedInvVar
)
{}
#endif // HL_CUDA_CUDNN_STUB_H_
paddle/cuda/include/stub/hl_cuda_stub.h
浏览文件 @
80c68d38
...
...
@@ -12,7 +12,6 @@ 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. */
#ifndef HL_CUDA_STUB_H_
#define HL_CUDA_STUB_H_
...
...
@@ -24,29 +23,25 @@ inline void hl_specify_devices_start(int *device, int number) {}
inline
void
hl_init
(
int
device
)
{}
inline
int
hl_get_cuda_lib_version
(
int
device
)
{
return
0
;
}
inline
int
hl_get_cuda_lib_version
(
int
device
)
{
return
0
;
}
inline
void
hl_fini
()
{}
inline
void
hl_set_sync_flag
(
bool
flag
)
{}
inline
bool
hl_get_sync_flag
()
{
return
false
;
}
inline
bool
hl_get_sync_flag
()
{
return
false
;
}
inline
int
hl_get_device_count
()
{
return
0
;
}
inline
int
hl_get_device_count
()
{
return
0
;
}
inline
void
hl_set_device
(
int
device
)
{}
inline
int
hl_get_device
()
{
return
0
;
}
inline
int
hl_get_device
()
{
return
0
;
}
inline
void
*
hl_malloc_device
(
size_t
size
)
{
return
NULL
;
}
inline
void
*
hl_malloc_device
(
size_t
size
)
{
return
NULL
;
}
inline
void
hl_free_mem_device
(
void
*
dest_d
)
{}
inline
void
*
hl_malloc_host
(
size_t
size
)
{
return
NULL
;
}
inline
void
*
hl_malloc_host
(
size_t
size
)
{
return
NULL
;
}
inline
void
hl_free_mem_host
(
void
*
dest_h
)
{}
...
...
@@ -64,7 +59,9 @@ inline void hl_rand(real *dest_d, size_t num) {}
inline
void
hl_srand
(
unsigned
int
seed
)
{}
inline
void
hl_memcpy_async
(
void
*
dst
,
void
*
src
,
size_t
size
,
inline
void
hl_memcpy_async
(
void
*
dst
,
void
*
src
,
size_t
size
,
hl_stream_t
stream
)
{}
inline
void
hl_stream_synchronize
(
hl_stream_t
stream
)
{}
...
...
@@ -83,11 +80,11 @@ inline void hl_stream_wait_event(hl_stream_t stream, hl_event_t event) {}
inline
void
hl_event_synchronize
(
hl_event_t
event
)
{}
inline
int
hl_get_device_last_error
()
{
return
0
;
}
inline
int
hl_get_device_last_error
()
{
return
0
;
}
inline
const
char
*
hl_get_device_error_string
()
{
return
NULL
;
}
inline
const
char
*
hl_get_device_error_string
()
{
return
NULL
;
}
inline
const
char
*
hl_get_device_error_string
(
size_t
err
)
{
return
NULL
;
}
inline
const
char
*
hl_get_device_error_string
(
size_t
err
)
{
return
NULL
;
}
inline
bool
hl_cuda_event_is_ready
(
hl_event_t
event
)
{
return
true
;
}
...
...
paddle/cuda/include/stub/hl_lstm_stub.h
浏览文件 @
80c68d38
...
...
@@ -12,7 +12,6 @@ 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. */
#ifndef HL_LSTM_STUB_H_
#define HL_LSTM_STUB_H_
...
...
paddle/cuda/include/stub/hl_matrix_stub.h
浏览文件 @
80c68d38
...
...
@@ -12,7 +12,6 @@ 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. */
#ifndef HL_MATRIX_STUB_H_
#define HL_MATRIX_STUB_H_
...
...
@@ -26,48 +25,30 @@ inline void hl_matrix_add(real* A_d,
real
alpha
,
real
beta
)
{}
inline
void
hl_matrix_softmax
(
real
*
A_d
,
real
*
C_d
,
int
dimM
,
int
dimN
)
{}
inline
void
hl_matrix_softmax
(
real
*
A_d
,
real
*
C_d
,
int
dimM
,
int
dimN
)
{}
inline
void
hl_sequence_softmax_forward
(
real
*
A_d
,
real
*
C_d
,
inline
void
hl_sequence_softmax_forward
(
real
*
A_d
,
real
*
C_d
,
const
int
*
index
,
int
numSequence
)
{}
inline
void
hl_matrix_softmax_derivative
(
real
*
grad_d
,
real
*
output_d
,
real
*
sftmaxSum_d
,
int
dimM
,
int
dimN
)
{}
inline
void
hl_matrix_classification_error
(
real
*
A_d
,
int
*
B_d
,
real
*
C_d
,
int
dimM
,
int
dimN
)
{}
inline
void
hl_matrix_cross_entropy
(
real
*
A_d
,
real
*
C_d
,
int
*
label_d
,
int
dimM
,
int
dimN
)
{}
inline
void
hl_matrix_cross_entropy_bp
(
real
*
grad_d
,
real
*
output_d
,
int
*
label_d
,
int
dimM
,
int
dimN
)
{}
inline
void
hl_matrix_multi_binary_cross_entropy
(
real
*
output
,
real
*
entropy
,
hl_sparse_matrix_s
mat
,
int
dimM
,
int
dimN
)
{}
inline
void
hl_matrix_multi_binary_cross_entropy_bp
(
real
*
output
,
real
*
grad
,
hl_sparse_matrix_s
mat
,
int
dimM
,
int
dimN
)
{}
inline
void
hl_matrix_softmax_derivative
(
real
*
grad_d
,
real
*
output_d
,
real
*
sftmaxSum_d
,
int
dimM
,
int
dimN
)
{}
inline
void
hl_matrix_classification_error
(
real
*
A_d
,
int
*
B_d
,
real
*
C_d
,
int
dimM
,
int
dimN
)
{}
inline
void
hl_matrix_cross_entropy
(
real
*
A_d
,
real
*
C_d
,
int
*
label_d
,
int
dimM
,
int
dimN
)
{}
inline
void
hl_matrix_cross_entropy_bp
(
real
*
grad_d
,
real
*
output_d
,
int
*
label_d
,
int
dimM
,
int
dimN
)
{}
inline
void
hl_matrix_multi_binary_cross_entropy
(
real
*
output
,
real
*
entropy
,
hl_sparse_matrix_s
mat
,
int
dimM
,
int
dimN
)
{}
inline
void
hl_matrix_multi_binary_cross_entropy_bp
(
real
*
output
,
real
*
grad
,
hl_sparse_matrix_s
mat
,
int
dimM
,
int
dimN
)
{}
inline
void
hl_matrix_zero_mem
(
real
*
data
,
int
num
)
{}
...
...
@@ -101,7 +82,6 @@ inline void hl_cossim(real* output,
int
input2_height
,
real
scale
)
{}
inline
void
hl_cossim_derivative
(
real
*
grad
,
real
*
output
,
real
*
prevOutX
,
...
...
paddle/cuda/include/stub/hl_sequence_stub.h
浏览文件 @
80c68d38
...
...
@@ -12,7 +12,6 @@ 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. */
#ifndef HL_SEQUENCE_STUB_H_
#define HL_SEQUENCE_STUB_H_
...
...
@@ -21,15 +20,12 @@ limitations under the License. */
inline
void
hl_max_sequence_forward
(
real
*
input
,
const
int
*
sequence
,
real
*
output
,
int
*
index
,
int
*
index
,
int
numSequences
,
int
dim
)
{}
inline
void
hl_max_sequence_backward
(
real
*
outputGrad
,
int
*
index
,
real
*
inputGrad
,
int
numSequences
,
int
dim
)
{}
inline
void
hl_max_sequence_backward
(
real
*
outputGrad
,
int
*
index
,
real
*
inputGrad
,
int
numSequences
,
int
dim
)
{}
inline
void
hl_context_projection_forward
(
real
*
input
,
const
int
*
sequence
,
...
...
@@ -60,16 +56,16 @@ inline void hl_context_projection_backward_weight(real* outputGrad,
int
contextStart
,
int
beginPad
)
{}
inline
void
hl_sequence2batch_copy
(
real
*
batch
,
real
*
sequence
,
const
int
*
batchIndex
,
inline
void
hl_sequence2batch_copy
(
real
*
batch
,
real
*
sequence
,
const
int
*
batchIndex
,
int
seqWidth
,
int
batchCount
,
bool
seq2batch
)
{}
inline
void
hl_sequence2batch_add
(
real
*
batch
,
real
*
sequence
,
int
*
batchIndex
,
inline
void
hl_sequence2batch_add
(
real
*
batch
,
real
*
sequence
,
int
*
batchIndex
,
int
seqWidth
,
int
batchCount
,
bool
seq2batch
)
{}
...
...
paddle/cuda/include/stub/hl_sparse_stub.h
浏览文件 @
80c68d38
...
...
@@ -12,7 +12,6 @@ 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. */
#ifndef HL_SPARSE_STUB_H_
#define HL_SPARSE_STUB_H_
...
...
@@ -20,7 +19,7 @@ limitations under the License. */
inline
void
hl_malloc_sparse_matrix
(
hl_sparse_matrix_s
*
A_d
,
hl_matrix_format_t
format
,
hl_matrix_value_t
value_type
,
hl_matrix_value_t
value_type
,
int
dimM
,
int
dimN
,
int
nnz
)
{}
...
...
@@ -28,20 +27,20 @@ inline void hl_malloc_sparse_matrix(hl_sparse_matrix_s *A_d,
inline
void
hl_free_sparse_matrix
(
hl_sparse_matrix_s
A_d
)
{}
inline
void
hl_construct_sparse_matrix
(
hl_sparse_matrix_s
*
A_d
,
void
*
dest_d
,
void
*
dest_d
,
size_t
size
,
hl_matrix_format_t
format
,
hl_matrix_value_t
value_type
,
hl_matrix_value_t
value_type
,
int
dimM
,
int
dimN
,
int
nnz
)
{}
inline
void
hl_construct_sparse_matrix
(
hl_sparse_matrix_s
*
A_d
,
real
*
value_d
,
int
*
rows_d
,
int
*
cols_d
,
real
*
value_d
,
int
*
rows_d
,
int
*
cols_d
,
hl_matrix_format_t
format
,
hl_matrix_value_t
value_type
,
hl_matrix_value_t
value_type
,
int
dimM
,
int
dimN
,
int
nnz
)
{}
...
...
@@ -87,10 +86,14 @@ inline void hl_matrix_csr_mul_dense(hl_sparse_matrix_s A_d,
inline
void
hl_matrix_csc_mul_dense
(
hl_sparse_matrix_s
A_d
,
hl_trans_op_t
transa
,
real
*
B_d
,
hl_trans_op_t
transb
,
real
*
B_d
,
hl_trans_op_t
transb
,
real
*
C_d
,
int
dimM
,
int
dimN
,
int
dimK
,
real
alpha
,
real
beta
)
{}
int
dimM
,
int
dimN
,
int
dimK
,
real
alpha
,
real
beta
)
{}
inline
void
hl_matrix_dense_mul_csc
(
real
*
A_d
,
hl_trans_op_t
transa
,
...
...
@@ -103,18 +106,27 @@ inline void hl_matrix_dense_mul_csc(real *A_d,
real
alpha
,
real
beta
)
{}
inline
void
hl_sparse_matrix_mul
(
real
*
A_d
,
hl_trans_op_t
transa
,
real
*
B_d
,
hl_trans_op_t
transb
,
inline
void
hl_sparse_matrix_mul
(
real
*
A_d
,
hl_trans_op_t
transa
,
real
*
B_d
,
hl_trans_op_t
transb
,
hl_sparse_matrix_s
C_d
,
int
dimM
,
int
dimN
,
int
dimK
,
real
alpha
,
real
beta
)
{}
int
dimM
,
int
dimN
,
int
dimK
,
real
alpha
,
real
beta
)
{}
inline
void
hl_matrix_dense_mul_csr
(
real
*
A_d
,
hl_trans_op_t
transa
,
inline
void
hl_matrix_dense_mul_csr
(
real
*
A_d
,
hl_trans_op_t
transa
,
hl_sparse_matrix_s
B_d
,
hl_trans_op_t
transb
,
real
*
C_d
,
int
dimM
,
int
dimN
,
int
dimK
,
real
alpha
,
real
beta
)
{}
int
dimM
,
int
dimN
,
int
dimK
,
real
alpha
,
real
beta
)
{}
inline
void
hl_memcpy_from_csc_matrix
(
real
*
csc_val
,
size_t
val_size
,
...
...
@@ -134,49 +146,39 @@ inline void hl_memcpy_from_csr_matrix(real *csr_val,
hl_sparse_matrix_s
csr_matrix
,
hl_stream_t
stream
)
{}
inline
void
hl_sparse_matrix_column_sum
(
real
*
A_d
,
hl_sparse_matrix_s
B_d
,
int
dimM
,
int
dimN
,
real
scale
)
{}
inline
void
hl_sparse_matrix_column_sum
(
real
*
A_d
,
hl_sparse_matrix_s
B_d
,
int
dimM
,
int
dimN
,
real
scale
)
{}
inline
void
hl_matrix_csr_column_sum
(
real
*
A_d
,
hl_sparse_matrix_s
B_d
,
int
dimM
,
int
dimN
,
real
scale
)
{}
inline
void
hl_matrix_csr_column_sum
(
real
*
A_d
,
hl_sparse_matrix_s
B_d
,
int
dimM
,
int
dimN
,
real
scale
)
{}
inline
void
hl_sparse_matrix_add_bias
(
hl_sparse_matrix_s
A_d
,
real
*
B_d
,
real
*
B_d
,
real
scale
)
{}
inline
void
hl_matrix_csr_add_bias
(
hl_sparse_matrix_s
A_d
,
real
*
B_d
,
real
*
B_d
,
real
scale
)
{}
inline
void
hl_sparse_matrix_add_dense
(
hl_sparse_matrix_s
A_d
,
real
*
B_d
,
real
*
B_d
,
int
dimM
,
int
dimN
,
real
alpha
,
real
beta
)
{}
inline
void
hl_matrix_csr_add_dense
(
hl_sparse_matrix_s
A_d
,
real
*
B_d
,
real
*
B_d
,
int
dimM
,
int
dimN
,
real
alpha
,
real
beta
)
{}
inline
int
*
hl_sparse_matrix_get_rows
(
hl_sparse_matrix_s
sMat
)
{
return
NULL
;
}
inline
int
*
hl_sparse_matrix_get_rows
(
hl_sparse_matrix_s
sMat
)
{
return
NULL
;
}
inline
int
*
hl_sparse_matrix_get_cols
(
hl_sparse_matrix_s
sMat
)
{
return
NULL
;
}
inline
int
*
hl_sparse_matrix_get_cols
(
hl_sparse_matrix_s
sMat
)
{
return
NULL
;
}
inline
real
*
hl_sparse_matrix_get_value
(
hl_sparse_matrix_s
sMat
)
{
inline
real
*
hl_sparse_matrix_get_value
(
hl_sparse_matrix_s
sMat
)
{
return
NULL
;
}
...
...
paddle/cuda/src/avx_mathfun.h
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/cuda/src/hl_avx_functions.cc
浏览文件 @
80c68d38
...
...
@@ -12,62 +12,58 @@ 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. */
#include <immintrin.h>
#include "hl_functions.h"
namespace
hppl
{
extern
__m256
exp
(
__m256
a
);
extern
__m256
exp
(
__m256
a
);
__m256
relu
(
const
__m256
a
)
{
__m256
tmp
=
_mm256_set1_ps
(
0.0
f
);
return
_mm256_max_ps
(
a
,
tmp
);
}
__m256
relu
(
const
__m256
a
)
{
__m256
tmp
=
_mm256_set1_ps
(
0.0
f
);
return
_mm256_max_ps
(
a
,
tmp
);
}
__m256
sigmoid
(
const
__m256
a
)
{
__m256
max
=
_mm256_set1_ps
(
SIGMOID_THRESHOLD_MAX
);
__m256
min
=
_mm256_set1_ps
(
SIGMOID_THRESHOLD_MIN
);
__m256
tmp
=
_mm256_max_ps
(
a
,
min
);
tmp
=
_mm256_min_ps
(
tmp
,
max
);
tmp
=
_mm256_sub_ps
(
_mm256_set1_ps
(
0.0
f
),
tmp
);
tmp
=
exp
(
tmp
);
tmp
=
_mm256_add_ps
(
_mm256_set1_ps
(
1.0
f
),
tmp
);
tmp
=
_mm256_div_ps
(
_mm256_set1_ps
(
1.0
f
),
tmp
);
return
tmp
;
}
__m256
sigmoid
(
const
__m256
a
)
{
__m256
max
=
_mm256_set1_ps
(
SIGMOID_THRESHOLD_MAX
);
__m256
min
=
_mm256_set1_ps
(
SIGMOID_THRESHOLD_MIN
);
__m256
tmp
=
_mm256_max_ps
(
a
,
min
);
tmp
=
_mm256_min_ps
(
tmp
,
max
);
tmp
=
_mm256_sub_ps
(
_mm256_set1_ps
(
0.0
f
),
tmp
);
tmp
=
exp
(
tmp
);
tmp
=
_mm256_add_ps
(
_mm256_set1_ps
(
1.0
f
),
tmp
);
tmp
=
_mm256_div_ps
(
_mm256_set1_ps
(
1.0
f
),
tmp
);
return
tmp
;
}
__m256
tanh
(
const
__m256
a
)
{
__m256
max
=
_mm256_set1_ps
(
EXP_MAX_INPUT
);
__m256
tmp
=
_mm256_mul_ps
(
_mm256_set1_ps
(
-
2.0
f
),
a
);
tmp
=
_mm256_min_ps
(
tmp
,
max
);
tmp
=
exp
(
tmp
);
return
_mm256_sub_ps
(
_mm256_div_ps
(
_mm256_set1_ps
(
2.0
f
),
_mm256_add_ps
(
_mm256_set1_ps
(
1.0
f
),
tmp
)),
_mm256_set1_ps
(
1.0
f
));
}
__m256
tanh
(
const
__m256
a
)
{
__m256
max
=
_mm256_set1_ps
(
EXP_MAX_INPUT
);
__m256
tmp
=
_mm256_mul_ps
(
_mm256_set1_ps
(
-
2.0
f
),
a
);
tmp
=
_mm256_min_ps
(
tmp
,
max
);
tmp
=
exp
(
tmp
);
return
_mm256_sub_ps
(
_mm256_div_ps
(
_mm256_set1_ps
(
2.0
f
),
_mm256_add_ps
(
_mm256_set1_ps
(
1.0
f
),
tmp
)
),
_mm256_set1_ps
(
1.0
f
));
}
__m256
linear
(
const
__m256
a
)
{
return
a
;
}
__m256
linear
(
const
__m256
a
)
{
return
a
;
}
__m256
relu
(
const
__m256
a
,
const
__m256
b
)
{
return
_mm256_mul_ps
(
a
,
__m256
relu
(
const
__m256
a
,
const
__m256
b
)
{
return
_mm256_mul_ps
(
a
,
_mm256_and_ps
(
_mm256_cmp_ps
(
b
,
_mm256_set1_ps
(
0.0
f
),
_CMP_GT_OS
),
_mm256_set1_ps
(
1.0
f
)));
}
_mm256_set1_ps
(
1.0
f
)));
}
__m256
sigmoid
(
const
__m256
a
,
const
__m256
b
)
{
return
_mm256_mul_ps
(
_mm256_mul_ps
(
a
,
b
),
_mm256_sub_ps
(
_mm256_set1_ps
(
1.0
f
),
b
));
}
__m256
sigmoid
(
const
__m256
a
,
const
__m256
b
)
{
return
_mm256_mul_ps
(
_mm256_mul_ps
(
a
,
b
),
_mm256_sub_ps
(
_mm256_set1_ps
(
1.0
f
),
b
));
}
__m256
tanh
(
const
__m256
a
,
const
__m256
b
)
{
return
_mm256_mul_ps
(
a
,
_mm256_sub_ps
(
_mm256_set1_ps
(
1.0
f
),
_mm256_mul_ps
(
b
,
b
)));
}
__m256
tanh
(
const
__m256
a
,
const
__m256
b
)
{
return
_mm256_mul_ps
(
a
,
_mm256_sub_ps
(
_mm256_set1_ps
(
1.0
f
),
_mm256_mul_ps
(
b
,
b
)));
}
__m256
linear
(
const
__m256
a
,
const
__m256
b
)
{
return
a
;
}
__m256
linear
(
const
__m256
a
,
const
__m256
b
)
{
return
a
;
}
}
// namespace hppl
paddle/cuda/src/hl_cpu_functions.cc
浏览文件 @
80c68d38
...
...
@@ -12,46 +12,33 @@ 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. */
#include <math.h>
#include "hl_functions.h"
namespace
hppl
{
real
relu
(
const
real
a
)
{
return
a
>
0.0
f
?
a
:
0.0
f
;
}
real
sigmoid
(
const
real
a
)
{
const
real
min
=
SIGMOID_THRESHOLD_MIN
;
const
real
max
=
SIGMOID_THRESHOLD_MAX
;
real
tmp
=
(
a
<
min
)
?
min
:
((
a
>
max
)
?
max
:
a
);
return
1.0
/
(
1.0
+
exp
(
-
tmp
));
}
real
tanh
(
const
real
a
)
{
real
tmp
=
-
2.0
*
a
;
tmp
=
(
tmp
>
EXP_MAX_INPUT
)
?
EXP_MAX_INPUT
:
tmp
;
return
(
2.0
/
(
1.0
+
exp
(
tmp
)))
-
1.0
;
}
real
linear
(
const
real
a
)
{
return
a
;
}
real
relu
(
const
real
a
,
const
real
b
)
{
return
a
*
(
b
>
0.0
f
?
1.0
f
:
0.0
f
);
}
real
sigmoid
(
const
real
a
,
const
real
b
)
{
return
a
*
b
*
(
1
-
b
);
}
real
tanh
(
const
real
a
,
const
real
b
)
{
return
a
*
(
1.0
f
-
b
*
b
);
}
real
linear
(
const
real
a
,
const
real
b
)
{
return
a
;
}
real
relu
(
const
real
a
)
{
return
a
>
0.0
f
?
a
:
0.0
f
;
}
real
sigmoid
(
const
real
a
)
{
const
real
min
=
SIGMOID_THRESHOLD_MIN
;
const
real
max
=
SIGMOID_THRESHOLD_MAX
;
real
tmp
=
(
a
<
min
)
?
min
:
((
a
>
max
)
?
max
:
a
);
return
1.0
/
(
1.0
+
exp
(
-
tmp
));
}
real
tanh
(
const
real
a
)
{
real
tmp
=
-
2.0
*
a
;
tmp
=
(
tmp
>
EXP_MAX_INPUT
)
?
EXP_MAX_INPUT
:
tmp
;
return
(
2.0
/
(
1.0
+
exp
(
tmp
)))
-
1.0
;
}
real
linear
(
const
real
a
)
{
return
a
;
}
real
relu
(
const
real
a
,
const
real
b
)
{
return
a
*
(
b
>
0.0
f
?
1.0
f
:
0.0
f
);
}
real
sigmoid
(
const
real
a
,
const
real
b
)
{
return
a
*
b
*
(
1
-
b
);
}
real
tanh
(
const
real
a
,
const
real
b
)
{
return
a
*
(
1.0
f
-
b
*
b
);
}
real
linear
(
const
real
a
,
const
real
b
)
{
return
a
;
}
}
// namespace hppl
paddle/cuda/src/hl_cuda_cublas.cc
浏览文件 @
80c68d38
...
...
@@ -12,7 +12,6 @@ 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. */
#include <sys/time.h>
#include <mutex>
#include "hl_cuda.h"
...
...
@@ -24,7 +23,7 @@ limitations under the License. */
namespace
dynload
{
std
::
once_flag
cublas_dso_flag
;
void
*
cublas_dso_handle
=
nullptr
;
void
*
cublas_dso_handle
=
nullptr
;
/**
* The following macro definition can generate structs
...
...
@@ -34,38 +33,32 @@ void* cublas_dso_handle = nullptr;
* note: default dynamic linked libs
*/
#ifdef PADDLE_USE_DSO
#define DYNAMIC_LOAD_CUBLAS_WRAP(__name) \
struct DynLoad__##__name { \
template <typename... Args> \
cublasStatus_t operator()(Args... args) { \
typedef cublasStatus_t (*cublasFunc)(Args...); \
std::call_once(cublas_dso_flag, GetCublasDsoHandle, \
&cublas_dso_handle); \
void* p_##__name = dlsym(cublas_dso_handle, #__name); \
return reinterpret_cast<cublasFunc>(p_##__name)(args...); \
} \
#define DYNAMIC_LOAD_CUBLAS_WRAP(__name) \
struct DynLoad__##__name { \
template <typename... Args> \
cublasStatus_t operator()(Args... args) { \
typedef cublasStatus_t (*cublasFunc)(Args...); \
std::call_once(cublas_dso_flag, GetCublasDsoHandle, &cublas_dso_handle); \
void *p_##__name = dlsym(cublas_dso_handle, #__name); \
return reinterpret_cast<cublasFunc>(p_##__name)(args...); \
} \
} __name; // struct DynLoad__##__name
#else
#define DYNAMIC_LOAD_CUBLAS_WRAP(__name)
\
struct DynLoad__##__name {
\
template <typename... Args>
\
cublasStatus_t operator()(Args... args) {
\
return __name(args...);
\
}
\
#define DYNAMIC_LOAD_CUBLAS_WRAP(__name) \
struct DynLoad__##__name {
\
template <typename... Args> \
cublasStatus_t operator()(Args... args) { \
return __name(args...); \
} \
} __name; // struct DynLoad__##__name
#endif
#define DYNAMIC_LOAD_CUBLAS_V2_WRAP(__name) \
DYNAMIC_LOAD_CUBLAS_WRAP(__name)
#define DYNAMIC_LOAD_CUBLAS_V2_WRAP(__name) DYNAMIC_LOAD_CUBLAS_WRAP(__name)
// include all needed cublas functions in HPPL
#define CUBLAS_BLAS_ROUTINE_EACH(__macro) \
__macro(cublasSgemv) \
__macro(cublasDgemv) \
__macro(cublasSgemm) \
__macro(cublasDgemm) \
__macro(cublasSgeam) \
__macro(cublasDgeam) \
#define CUBLAS_BLAS_ROUTINE_EACH(__macro) \
__macro(cublasSgemv) __macro(cublasDgemv) __macro(cublasSgemm) \
__macro(cublasDgemm) __macro(cublasSgeam) __macro(cublasDgeam)
DYNAMIC_LOAD_CUBLAS_V2_WRAP
(
cublasCreate
)
DYNAMIC_LOAD_CUBLAS_V2_WRAP
(
cublasDestroy
)
...
...
@@ -88,41 +81,40 @@ CUBLAS_BLAS_ROUTINE_EACH(DYNAMIC_LOAD_CUBLAS_V2_WRAP)
}
/* namespace dynload */
#ifndef PADDLE_TYPE_DOUBLE
#define
CUBLAS_GEAM
dynload::cublasSgeam
#define
CUBLAS_GEMV
dynload::cublasSgemv
#define
CUBLAS_GEMM
dynload::cublasSgemm
#define
CUBLAS_GETRF
dynload::cublasSgetrfBatched
#define
CUBLAS_GETRI
dynload::cublasSgetriBatched
#define
CUBLAS_GEAM
dynload::cublasSgeam
#define
CUBLAS_GEMV
dynload::cublasSgemv
#define
CUBLAS_GEMM
dynload::cublasSgemm
#define
CUBLAS_GETRF
dynload::cublasSgetrfBatched
#define
CUBLAS_GETRI
dynload::cublasSgetriBatched
#else
#define
CUBLAS_GEAM
dynload::cublasDgeam
#define
CUBLAS_GEMV
dynload::cublasDgemv
#define
CUBLAS_GEMM
dynload::cublasDgemm
#define
CUBLAS_GETRF
dynload::cublasDgetrfBatched
#define
CUBLAS_GETRI
dynload::cublasDgetriBatched
#define
CUBLAS_GEAM
dynload::cublasDgeam
#define
CUBLAS_GEMV
dynload::cublasDgemv
#define
CUBLAS_GEMM
dynload::cublasDgemm
#define
CUBLAS_GETRF
dynload::cublasDgetrfBatched
#define
CUBLAS_GETRI
dynload::cublasDgetriBatched
#endif
const
char
*
hl_cublas_get_error_string
(
cublasStatus_t
status
)
{
const
char
*
hl_cublas_get_error_string
(
cublasStatus_t
status
)
{
switch
(
status
)
{
case
CUBLAS_STATUS_NOT_INITIALIZED
:
return
"[cublas status]: not initialized"
;
case
CUBLAS_STATUS_ALLOC_FAILED
:
return
"[cublas status]: allocate failed"
;
case
CUBLAS_STATUS_INVALID_VALUE
:
return
"[cublas status]: invalid value"
;
case
CUBLAS_STATUS_ARCH_MISMATCH
:
return
"[cublas status]: arch mismatch"
;
case
CUBLAS_STATUS_MAPPING_ERROR
:
return
"[cublas status]: mapping error"
;
case
CUBLAS_STATUS_EXECUTION_FAILED
:
return
"[cublas status]: execution failed"
;
case
CUBLAS_STATUS_INTERNAL_ERROR
:
return
"[cublas status]: internal error"
;
case
CUBLAS_STATUS_SUCCESS
:
return
"[cublas status]: success"
;
default:
return
"[cublas status]: unknown error"
;
case
CUBLAS_STATUS_NOT_INITIALIZED
:
return
"[cublas status]: not initialized"
;
case
CUBLAS_STATUS_ALLOC_FAILED
:
return
"[cublas status]: allocate failed"
;
case
CUBLAS_STATUS_INVALID_VALUE
:
return
"[cublas status]: invalid value"
;
case
CUBLAS_STATUS_ARCH_MISMATCH
:
return
"[cublas status]: arch mismatch"
;
case
CUBLAS_STATUS_MAPPING_ERROR
:
return
"[cublas status]: mapping error"
;
case
CUBLAS_STATUS_EXECUTION_FAILED
:
return
"[cublas status]: execution failed"
;
case
CUBLAS_STATUS_INTERNAL_ERROR
:
return
"[cublas status]: internal error"
;
case
CUBLAS_STATUS_SUCCESS
:
return
"[cublas status]: success"
;
default:
return
"[cublas status]: unknown error"
;
}
}
...
...
@@ -131,27 +123,21 @@ const char* hl_cublas_get_error_string(cublasStatus_t status) {
* support << operator for more details error info.
*/
cublasStatus_t
g_cublasStat
;
#define CHECK_CUBLAS(cublas_func) \
g_cublasStat = cublas_func; \
CHECK_EQ(CUBLAS_STATUS_SUCCESS, g_cublasStat) \
<< "Cublas Error: " \
<< hl_cublas_get_error_string(g_cublasStat) \
<< " "
#define CHECK_CUBLAS(cublas_func) \
g_cublasStat = cublas_func; \
CHECK_EQ(CUBLAS_STATUS_SUCCESS, g_cublasStat) \
<< "Cublas Error: " << hl_cublas_get_error_string(g_cublasStat) << " "
void
hl_cublas_init
(
cublasHandle_t
*
cublas_handle
,
cudaStream_t
stream
)
{
CHECK_CUBLAS
(
dynload
::
cublasCreate
(
cublas_handle
))
<<
"[cublas init] Cublas create handle faild!"
;
<<
"[cublas init] Cublas create handle faild!"
;
CHECK_CUBLAS
(
dynload
::
cublasSetStream
(
*
cublas_handle
,
stream
))
<<
"[cublas init] Cublas set stream faild!"
;
<<
"[cublas init] Cublas set stream faild!"
;
}
void
hl_matrix_transpose
(
real
*
A_d
,
real
*
C_d
,
int
dimM
,
int
dimN
,
int
lda
,
int
ldc
)
{
void
hl_matrix_transpose
(
real
*
A_d
,
real
*
C_d
,
int
dimM
,
int
dimN
,
int
lda
,
int
ldc
)
{
real
alpha
=
1.0
;
real
beta
=
0.0
;
...
...
@@ -159,11 +145,18 @@ void hl_matrix_transpose(real *A_d,
CHECK_NOTNULL
(
C_d
);
CHECK_CUBLAS
(
CUBLAS_GEAM
(
t_resource
.
handle
,
CUBLAS_OP_T
,
CUBLAS_OP_N
,
dimM
,
dimN
,
&
alpha
,
A_d
,
lda
,
&
beta
,
nullptr
,
dimM
,
C_d
,
ldc
));
CUBLAS_OP_T
,
CUBLAS_OP_N
,
dimM
,
dimN
,
&
alpha
,
A_d
,
lda
,
&
beta
,
nullptr
,
dimM
,
C_d
,
ldc
));
CHECK_SYNC
(
"hl_matrix_transpose failed"
);
}
...
...
@@ -188,13 +181,13 @@ void hl_matrix_inverse(real *A_d, real *C_d, int dimN, int lda, int ldc) {
small-sized matrices. There may be a better way to reconstruct
the API for better performance.
*/
CHECK_CUBLAS
(
CUBLAS_GETRF
(
t_resource
.
handle
,
dimN
,
inout_d
,
lda
,
pivot_d
,
info_d
,
1
));
CHECK_CUBLAS
(
CUBLAS_GETRF
(
t_resource
.
handle
,
dimN
,
inout_d
,
lda
,
pivot_d
,
info_d
,
1
));
int
info_h
;
hl_memcpy
(
&
info_h
,
info_d
,
sizeof
(
int
));
if
(
info_h
!=
0
)
{
LOG
(
FATAL
)
<<
"Factorization of matrix failed: matrix may be singular.
\n
"
;
LOG
(
FATAL
)
<<
"Factorization of matrix failed: matrix may be singular.
\n
"
;
}
/* Step 2: Compute the inverse of the matrix given its LU decomposition */
...
...
@@ -203,12 +196,18 @@ void hl_matrix_inverse(real *A_d, real *C_d, int dimN, int lda, int ldc) {
hl_memcpy
(
out_d
,
out_h
,
sizeof
(
real
*
));
CHECK_CUBLAS
(
CUBLAS_GETRI
(
t_resource
.
handle
,
dimN
,
(
const
real
**
)
inout_d
,
lda
,
pivot_d
,
out_d
,
ldc
,
info_d
,
1
));
dimN
,
(
const
real
**
)
inout_d
,
lda
,
pivot_d
,
out_d
,
ldc
,
info_d
,
1
));
hl_memcpy
(
&
info_h
,
info_d
,
sizeof
(
int
));
if
(
info_h
!=
0
)
{
LOG
(
FATAL
)
<<
"Inversion of matrix failed: matrix may be singular.
\n
"
;
LOG
(
FATAL
)
<<
"Inversion of matrix failed: matrix may be singular.
\n
"
;
}
hl_free_mem_device
(
inout_d
);
...
...
@@ -218,12 +217,19 @@ void hl_matrix_inverse(real *A_d, real *C_d, int dimN, int lda, int ldc) {
CHECK_SYNC
(
"hl_matrix_inverse failed"
);
}
void
hl_matrix_mul
(
real
*
A_d
,
hl_trans_op_t
transa
,
real
*
B_d
,
hl_trans_op_t
transb
,
void
hl_matrix_mul
(
real
*
A_d
,
hl_trans_op_t
transa
,
real
*
B_d
,
hl_trans_op_t
transb
,
real
*
C_d
,
int
dimM
,
int
dimN
,
int
dimK
,
real
alpha
,
real
beta
,
int
lda
,
int
ldb
,
int
ldc
)
{
int
dimM
,
int
dimN
,
int
dimK
,
real
alpha
,
real
beta
,
int
lda
,
int
ldb
,
int
ldc
)
{
CHECK_NOTNULL
(
A_d
);
CHECK_NOTNULL
(
B_d
);
CHECK_NOTNULL
(
C_d
);
...
...
@@ -231,8 +237,8 @@ void hl_matrix_mul(real *A_d, hl_trans_op_t transa,
if
(
dimN
==
1
&&
dimM
!=
1
&&
dimK
!=
1
&&
transb
==
HPPL_OP_N
)
{
int
m
=
(
transa
==
HPPL_OP_N
)
?
dimM
:
dimK
;
int
n
=
(
transa
==
HPPL_OP_N
)
?
dimK
:
dimM
;
hl_matrix_mul_vector
(
A_d
,
transa
,
B_d
,
C_d
,
m
,
n
,
alpha
,
beta
,
lda
,
ldb
,
ldc
);
hl_matrix_mul_vector
(
A_d
,
transa
,
B_d
,
C_d
,
m
,
n
,
alpha
,
beta
,
lda
,
ldb
,
ldc
);
return
;
}
...
...
@@ -240,8 +246,7 @@ void hl_matrix_mul(real *A_d, hl_trans_op_t transa,
int
m
=
(
transb
==
HPPL_OP_N
)
?
dimK
:
dimN
;
int
n
=
(
transb
==
HPPL_OP_N
)
?
dimN
:
dimK
;
hl_trans_op_t
trans
=
(
transb
==
HPPL_OP_N
)
?
HPPL_OP_T
:
HPPL_OP_N
;
hl_matrix_mul_vector
(
B_d
,
trans
,
A_d
,
C_d
,
m
,
n
,
alpha
,
beta
,
ldb
,
1
,
1
);
hl_matrix_mul_vector
(
B_d
,
trans
,
A_d
,
C_d
,
m
,
n
,
alpha
,
beta
,
ldb
,
1
,
1
);
return
;
}
...
...
@@ -250,26 +255,47 @@ void hl_matrix_mul(real *A_d, hl_trans_op_t transa,
stat
=
CUBLAS_GEMM
(
t_resource
.
handle
,
CUBLAS_OP_N
,
CUBLAS_OP_N
,
dimN
,
dimM
,
dimK
,
&
alpha
,
B_d
,
ldb
,
A_d
,
lda
,
&
beta
,
C_d
,
ldc
);
dimN
,
dimM
,
dimK
,
&
alpha
,
B_d
,
ldb
,
A_d
,
lda
,
&
beta
,
C_d
,
ldc
);
}
else
if
((
HPPL_OP_T
==
transa
)
&&
(
HPPL_OP_N
==
transb
))
{
stat
=
CUBLAS_GEMM
(
t_resource
.
handle
,
CUBLAS_OP_N
,
CUBLAS_OP_T
,
dimN
,
dimM
,
dimK
,
&
alpha
,
B_d
,
ldb
,
A_d
,
lda
,
&
beta
,
C_d
,
ldc
);
dimN
,
dimM
,
dimK
,
&
alpha
,
B_d
,
ldb
,
A_d
,
lda
,
&
beta
,
C_d
,
ldc
);
}
else
if
((
HPPL_OP_N
==
transa
)
&&
(
HPPL_OP_T
==
transb
))
{
stat
=
CUBLAS_GEMM
(
t_resource
.
handle
,
CUBLAS_OP_T
,
CUBLAS_OP_N
,
dimN
,
dimM
,
dimK
,
&
alpha
,
B_d
,
ldb
,
A_d
,
lda
,
&
beta
,
C_d
,
ldc
);
dimN
,
dimM
,
dimK
,
&
alpha
,
B_d
,
ldb
,
A_d
,
lda
,
&
beta
,
C_d
,
ldc
);
}
else
{
LOG
(
FATAL
)
<<
"parameter transa error!"
;
}
...
...
@@ -277,24 +303,46 @@ void hl_matrix_mul(real *A_d, hl_trans_op_t transa,
CHECK_SYNC
(
"hl_matrix_mul failed"
);
}
void
hl_matrix_mul
(
real
*
A_d
,
hl_trans_op_t
transa
,
real
*
B_d
,
hl_trans_op_t
transb
,
void
hl_matrix_mul
(
real
*
A_d
,
hl_trans_op_t
transa
,
real
*
B_d
,
hl_trans_op_t
transb
,
real
*
C_d
,
int
dimM
,
int
dimN
,
int
dimK
,
real
alpha
,
real
beta
)
{
int
dimM
,
int
dimN
,
int
dimK
,
real
alpha
,
real
beta
)
{
int
lda
=
(
HPPL_OP_N
==
transa
)
?
dimK
:
dimM
;
int
ldb
=
(
HPPL_OP_N
==
transb
)
?
dimN
:
dimK
;
int
ldc
=
dimN
;
hl_matrix_mul
(
A_d
,
transa
,
B_d
,
transb
,
C_d
,
dimM
,
dimN
,
dimK
,
alpha
,
beta
,
lda
,
ldb
,
ldc
);
hl_matrix_mul
(
A_d
,
transa
,
B_d
,
transb
,
C_d
,
dimM
,
dimN
,
dimK
,
alpha
,
beta
,
lda
,
ldb
,
ldc
);
}
void
hl_matrix_mul_vector
(
real
*
A_d
,
hl_trans_op_t
trans
,
real
*
B_d
,
real
*
C_d
,
int
dimM
,
int
dimN
,
real
alpha
,
real
beta
,
int
lda
,
int
incb
,
int
incc
)
{
void
hl_matrix_mul_vector
(
real
*
A_d
,
hl_trans_op_t
trans
,
real
*
B_d
,
real
*
C_d
,
int
dimM
,
int
dimN
,
real
alpha
,
real
beta
,
int
lda
,
int
incb
,
int
incc
)
{
CHECK_NOTNULL
(
A_d
);
CHECK_NOTNULL
(
B_d
);
CHECK_NOTNULL
(
C_d
);
...
...
@@ -303,21 +351,29 @@ void hl_matrix_mul_vector(real *A_d, hl_trans_op_t trans,
if
(
HPPL_OP_N
==
trans
)
{
stat
=
CUBLAS_GEMV
(
t_resource
.
handle
,
CUBLAS_OP_T
,
dimN
,
dimM
,
dimN
,
dimM
,
&
alpha
,
A_d
,
lda
,
B_d
,
incb
,
A_d
,
lda
,
B_d
,
incb
,
&
beta
,
C_d
,
incc
);
C_d
,
incc
);
}
else
if
(
HPPL_OP_T
==
trans
)
{
stat
=
CUBLAS_GEMV
(
t_resource
.
handle
,
CUBLAS_OP_N
,
dimN
,
dimM
,
dimN
,
dimM
,
&
alpha
,
A_d
,
lda
,
B_d
,
incb
,
A_d
,
lda
,
B_d
,
incb
,
&
beta
,
C_d
,
incc
);
C_d
,
incc
);
}
else
{
LOG
(
FATAL
)
<<
"parameter transa error!"
;
}
...
...
@@ -326,10 +382,14 @@ void hl_matrix_mul_vector(real *A_d, hl_trans_op_t trans,
CHECK_SYNC
(
"hl_matrix_mul_vector"
);
}
void
hl_matrix_mul_vector
(
real
*
A_d
,
hl_trans_op_t
trans
,
real
*
B_d
,
real
*
C_d
,
int
dimM
,
int
dimN
,
real
alpha
,
real
beta
)
{
hl_matrix_mul_vector
(
A_d
,
trans
,
B_d
,
C_d
,
dimM
,
dimN
,
alpha
,
beta
,
dimN
,
1
,
1
);
void
hl_matrix_mul_vector
(
real
*
A_d
,
hl_trans_op_t
trans
,
real
*
B_d
,
real
*
C_d
,
int
dimM
,
int
dimN
,
real
alpha
,
real
beta
)
{
hl_matrix_mul_vector
(
A_d
,
trans
,
B_d
,
C_d
,
dimM
,
dimN
,
alpha
,
beta
,
dimN
,
1
,
1
);
}
paddle/cuda/src/hl_cuda_cudnn.cc
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/cuda/src/hl_cuda_device.cc
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/cuda/src/hl_cudart_wrap.cc
浏览文件 @
80c68d38
...
...
@@ -12,7 +12,6 @@ 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. */
#ifdef PADDLE_USE_DSO
#include <mutex>
...
...
@@ -29,48 +28,46 @@ limitations under the License. */
namespace
dynload
{
extern
std
::
once_flag
cudart_dso_flag
;
extern
void
*
cudart_dso_handle
;
extern
void
*
cudart_dso_handle
;
/**
* The following macro definition can generate structs
* (for each function) to dynamic load cuda routine
* via operator overloading.
**/
#define DYNAMIC_LOAD_CUDART_WRAP(__name, __type) \
struct DynLoad__##__name { \
template <typename... Args> \
__type operator()(Args... args) { \
typedef __type (*cudartFunc)(Args...); \
std::call_once(cudart_dso_flag, GetCudartDsoHandle, \
&cudart_dso_handle); \
void* p_##__name = dlsym(cudart_dso_handle, #__name); \
return reinterpret_cast<cudartFunc>(p_##__name)(args...); \
} \
} __name;
/* struct DynLoad__##__name */
#define DYNAMIC_LOAD_CUDART_WRAP(__name, __type) \
struct DynLoad__##__name { \
template <typename... Args> \
__type operator()(Args... args) { \
typedef __type (*cudartFunc)(Args...); \
std::call_once(cudart_dso_flag, GetCudartDsoHandle, &cudart_dso_handle); \
void *p_##__name = dlsym(cudart_dso_handle, #__name); \
return reinterpret_cast<cudartFunc>(p_##__name)(args...); \
} \
} __name;
/* struct DynLoad__##__name */
/* include all needed cuda functions in HPPL */
#define CUDA_ROUTINE_EACH(__macro) \
__macro(cudaLaunch, cudaError_t) \
__macro(cudaSetupArgument, cudaError_t) \
__macro(cudaConfigureCall, cudaError_t) \
__macro(__cudaRegisterFatBinary, void**) \
__macro(__cudaUnregisterFatBinary, void) \
__macro(__cudaRegisterFunction, void) \
__macro(__cudaRegisterVar, void) \
__macro(__cudaRegisterManagedVar, void) \
__macro(__cudaInitModule, char) \
__macro(__cudaRegisterTexture, void) \
__macro(__cudaRegisterSurface, void)
#define CUDA_ROUTINE_EACH(__macro) \
__macro(cudaLaunch, cudaError_t) __macro(cudaSetupArgument, cudaError_t) \
__macro(cudaConfigureCall, cudaError_t) \
__macro(__cudaRegisterFatBinary, void **) \
__macro(__cudaUnregisterFatBinary, void) \
__macro(__cudaRegisterFunction, void) \
__macro(__cudaRegisterVar, void) \
__macro(__cudaRegisterManagedVar, void) \
__macro(__cudaInitModule, char) \
__macro(__cudaRegisterTexture, void) \
__macro(__cudaRegisterSurface, void)
CUDA_ROUTINE_EACH
(
DYNAMIC_LOAD_CUDART_WRAP
)
#if CUDART_VERSION >= 7000
DYNAMIC_LOAD_CUDART_WRAP
(
cudaLaunchKernel
,
cudaError_t
)
DYNAMIC_LOAD_CUDART_WRAP
(
cudaLaunchKernel
,
cudaError_t
)
#endif
#undef CUDA_ROUNTINE_EACH
}
/* namespace dynload */
}
/* namespace dynload */
#if CUDART_VERSION >= 7000
__host__
cudaError_t
CUDARTAPI
cudaLaunchKernel
(
const
void
*
func
,
...
...
@@ -79,12 +76,11 @@ __host__ cudaError_t CUDARTAPI cudaLaunchKernel(const void *func,
void
**
args
,
size_t
sharedMem
,
cudaStream_t
stream
)
{
return
dynload
::
cudaLaunchKernel
(
func
,
gridDim
,
blockDim
,
args
,
sharedMem
,
stream
);
return
dynload
::
cudaLaunchKernel
(
func
,
gridDim
,
blockDim
,
args
,
sharedMem
,
stream
);
}
#endif
/* CUDART_VERSION >= 7000 */
__host__
cudaError_t
CUDARTAPI
cudaLaunch
(
const
void
*
func
)
{
return
dynload
::
cudaLaunch
(
func
);
}
...
...
@@ -99,13 +95,12 @@ __host__ cudaError_t CUDARTAPI cudaConfigureCall(dim3 gridDim,
dim3
blockDim
,
size_t
sharedMem
,
cudaStream_t
stream
)
{
return
dynload
::
cudaConfigureCall
(
gridDim
,
blockDim
,
sharedMem
,
stream
);
return
dynload
::
cudaConfigureCall
(
gridDim
,
blockDim
,
sharedMem
,
stream
);
}
extern
"C"
{
void
**
CUDARTAPI
__cudaRegisterFatBinary
(
void
*
fatCubin
)
{
void
**
CUDARTAPI
__cudaRegisterFatBinary
(
void
*
fatCubin
)
{
return
dynload
::
__cudaRegisterFatBinary
(
fatCubin
);
}
...
...
@@ -113,86 +108,87 @@ void CUDARTAPI __cudaUnregisterFatBinary(void **fatCubinHandle) {
return
dynload
::
__cudaUnregisterFatBinary
(
fatCubinHandle
);
}
void
CUDARTAPI
__cudaRegisterFunction
(
void
**
fatCubinHandle
,
const
char
*
hostFun
,
char
*
deviceFun
,
const
char
*
deviceName
,
int
thread_limit
,
uint3
*
tid
,
uint3
*
bid
,
dim3
*
bDim
,
dim3
*
gDim
,
int
*
wSize
)
{
return
dynload
::
__cudaRegisterFunction
(
fatCubinHandle
,
hostFun
,
deviceFun
,
deviceName
,
thread_limit
,
tid
,
bid
,
bDim
,
gDim
,
wSize
);
void
CUDARTAPI
__cudaRegisterFunction
(
void
**
fatCubinHandle
,
const
char
*
hostFun
,
char
*
deviceFun
,
const
char
*
deviceName
,
int
thread_limit
,
uint3
*
tid
,
uint3
*
bid
,
dim3
*
bDim
,
dim3
*
gDim
,
int
*
wSize
)
{
return
dynload
::
__cudaRegisterFunction
(
fatCubinHandle
,
hostFun
,
deviceFun
,
deviceName
,
thread_limit
,
tid
,
bid
,
bDim
,
gDim
,
wSize
);
}
void
CUDARTAPI
__cudaRegisterVar
(
void
**
fatCubinHandle
,
char
*
hostVar
,
char
*
deviceAddress
,
const
char
*
deviceName
,
int
ext
,
int
size
,
int
constant
,
int
global
)
{
return
dynload
::
__cudaRegisterVar
(
fatCubinHandle
,
hostVar
,
deviceAddress
,
deviceName
,
ext
,
size
,
constant
,
global
);
void
CUDARTAPI
__cudaRegisterVar
(
void
**
fatCubinHandle
,
char
*
hostVar
,
char
*
deviceAddress
,
const
char
*
deviceName
,
int
ext
,
int
size
,
int
constant
,
int
global
)
{
return
dynload
::
__cudaRegisterVar
(
fatCubinHandle
,
hostVar
,
deviceAddress
,
deviceName
,
ext
,
size
,
constant
,
global
);
}
extern
void
CUDARTAPI
__cudaRegisterManagedVar
(
void
**
fatCubinHandle
,
void
**
hostVarPtrAddress
,
char
*
deviceAddress
,
const
char
*
deviceName
,
int
ext
,
int
size
,
int
constant
,
int
global
)
{
return
dynload
::
__cudaRegisterManagedVar
(
fatCubinHandle
,
hostVarPtrAddress
,
deviceAddress
,
deviceName
,
ext
,
size
,
constant
,
global
);
extern
void
CUDARTAPI
__cudaRegisterManagedVar
(
void
**
fatCubinHandle
,
void
**
hostVarPtrAddress
,
char
*
deviceAddress
,
const
char
*
deviceName
,
int
ext
,
int
size
,
int
constant
,
int
global
)
{
return
dynload
::
__cudaRegisterManagedVar
(
fatCubinHandle
,
hostVarPtrAddress
,
deviceAddress
,
deviceName
,
ext
,
size
,
constant
,
global
);
}
char
CUDARTAPI
__cudaInitModule
(
void
**
fatCubinHandle
)
{
char
CUDARTAPI
__cudaInitModule
(
void
**
fatCubinHandle
)
{
return
dynload
::
__cudaInitModule
(
fatCubinHandle
);
}
void
CUDARTAPI
__cudaRegisterTexture
(
void
**
fatCubinHandle
,
const
struct
textureReference
*
hostVar
,
const
void
**
deviceAddress
,
const
char
*
deviceName
,
int
dim
,
int
norm
,
int
ext
)
{
void
CUDARTAPI
__cudaRegisterTexture
(
void
**
fatCubinHandle
,
const
struct
textureReference
*
hostVar
,
const
void
**
deviceAddress
,
const
char
*
deviceName
,
int
dim
,
int
norm
,
int
ext
)
{
return
dynload
::
__cudaRegisterTexture
(
fatCubinHandle
,
hostVar
,
deviceAddress
,
deviceName
,
dim
,
norm
,
ext
);
fatCubinHandle
,
hostVar
,
deviceAddress
,
deviceName
,
dim
,
norm
,
ext
);
}
void
CUDARTAPI
__cudaRegisterSurface
(
void
**
fatCubinHandle
,
const
struct
surfaceReference
*
hostVar
,
const
void
**
deviceAddress
,
const
char
*
deviceName
,
int
dim
,
int
ext
)
{
void
CUDARTAPI
__cudaRegisterSurface
(
void
**
fatCubinHandle
,
const
struct
surfaceReference
*
hostVar
,
const
void
**
deviceAddress
,
const
char
*
deviceName
,
int
dim
,
int
ext
)
{
return
dynload
::
__cudaRegisterSurface
(
fatCubinHandle
,
hostVar
,
deviceAddress
,
deviceName
,
dim
,
ext
);
fatCubinHandle
,
hostVar
,
deviceAddress
,
deviceName
,
dim
,
ext
);
}
}
/* extern "C" */
...
...
paddle/cuda/src/hl_math.cc
浏览文件 @
80c68d38
...
...
@@ -12,24 +12,15 @@ 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. */
#include "avx_mathfun.h"
namespace
hppl
{
__m256
exp
(
__m256
a
)
{
return
exp256_ps
(
a
);
}
__m256
exp
(
__m256
a
)
{
return
exp256_ps
(
a
);
}
__m256
log
(
__m256
a
)
{
return
log256_ps
(
a
);
}
__m256
log
(
__m256
a
)
{
return
log256_ps
(
a
);
}
__m256
sin
(
__m256
a
)
{
return
sin256_ps
(
a
);
}
__m256
sin
(
__m256
a
)
{
return
sin256_ps
(
a
);
}
__m256
cos
(
__m256
a
)
{
return
cos256_ps
(
a
);
}
__m256
cos
(
__m256
a
)
{
return
cos256_ps
(
a
);
}
}
// namespace hppl
paddle/cuda/src/hl_time.cc
浏览文件 @
80c68d38
...
...
@@ -12,7 +12,6 @@ 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. */
#include <chrono>
#include <stdlib.h>
#include <iostream>
...
...
@@ -21,8 +20,7 @@ limitations under the License. */
using
std
::
chrono
::
high_resolution_clock
;
int64_t
getCurrentTimeStick
()
{
high_resolution_clock
::
time_point
tp
=
high_resolution_clock
::
now
();
high_resolution_clock
::
duration
dtn
=
tp
.
time_since_epoch
();
return
dtn
.
count
();
high_resolution_clock
::
time_point
tp
=
high_resolution_clock
::
now
();
high_resolution_clock
::
duration
dtn
=
tp
.
time_since_epoch
();
return
dtn
.
count
();
}
paddle/gserver/activations/ActivationFunction.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/activations/ActivationFunction.h
浏览文件 @
80c68d38
...
...
@@ -12,7 +12,6 @@ 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. */
#pragma once
#include <string>
#include <vector>
...
...
paddle/gserver/dataproviders/DataProvider.cpp
浏览文件 @
80c68d38
...
...
@@ -12,7 +12,6 @@ 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. */
#include "DataProvider.h"
#include "paddle/utils/Util.h"
...
...
@@ -57,7 +56,7 @@ void BufferBatch::clone(DataBatch* srcBatch, bool useGpu) {
}
}
DoubleBuffer
::
DoubleBuffer
(
DataProvider
*
dataPool
,
DoubleBuffer
::
DoubleBuffer
(
DataProvider
*
dataPool
,
bool
useGpu
,
int64_t
batchSize
)
{
batchSize_
=
batchSize
;
...
...
@@ -155,7 +154,7 @@ void DoubleBuffer::startAsyncLoad() {
}
ClassRegistrar
<
DataProvider
,
DataConfig
,
ModelConfig
,
bool
>
DataProvider
::
registrar_
;
DataProvider
::
registrar_
;
DataProvider
*
DataProvider
::
create
(
const
DataConfig
&
config
,
const
ModelConfig
&
modelConfig
,
...
...
@@ -182,7 +181,8 @@ int64_t DataProvider::getNextBatch(int64_t size, DataBatch* batch) {
for
(
int
i
=
0
;
i
<
config_
.
constant_slots_size
();
++
i
)
{
MemoryHandlePtr
handle
=
constantSlots
[
i
]
?
constantSlots
[
i
]
->
getMemoryHandle
()
:
nullptr
;
Matrix
::
resizeOrCreate
(
constantSlots
[
i
],
batchSize
,
Matrix
::
resizeOrCreate
(
constantSlots
[
i
],
batchSize
,
1
,
// = width
false
,
// = trans
useGpu_
);
// = useGpu
...
...
@@ -216,7 +216,8 @@ void DataProvider::initAsyncLoader() {
}
SimpleDataProviderBase
::
SimpleDataProviderBase
(
const
DataConfig
&
config
,
bool
useGpu
,
bool
withInfo
)
bool
useGpu
,
bool
withInfo
)
:
DataProvider
(
config
,
useGpu
)
{
/* initialize the size of a sample, and the buffer */
sampleDim_
=
config_
.
feat_dim
()
*
(
2
*
config_
.
context_len
()
+
1
);
...
...
@@ -337,7 +338,8 @@ int64_t SimpleDataProviderBase::fillBuffer() {
sampleNumInBuf_
=
n
+
fillBufferImp
(
hInputDataBuf_
->
getData
()
+
n
*
sampleDim_
,
hInputLabelBuf_
->
getData
()
+
n
,
hInputInfoBuf_
->
getData
()
+
n
,
bufferCapacity_
-
n
);
hInputInfoBuf_
->
getData
()
+
n
,
bufferCapacity_
-
n
);
/* for stachastic gradient training */
if
(
!
skipShuffle_
)
{
...
...
@@ -357,11 +359,14 @@ SimpleDataProvider::SimpleDataProvider(const DataConfig& config, bool useGpu)
SimpleDataProvider
::~
SimpleDataProvider
()
{}
int64_t
SimpleDataProvider
::
fillBufferImp
(
real
*
data
,
int
*
label
,
int
*
info
,
int64_t
SimpleDataProvider
::
fillBufferImp
(
real
*
data
,
int
*
label
,
int
*
info
,
int64_t
size
)
{
(
void
)
info
;
int64_t
n
=
std
::
min
<
int64_t
>
(
labels_
.
size
()
-
currentSampleIndex_
,
size
);
memcpy
(
data
,
&
data_
[
currentSampleIndex_
*
sampleDim_
],
memcpy
(
data
,
&
data_
[
currentSampleIndex_
*
sampleDim_
],
n
*
sampleDim_
*
sizeof
(
real
));
memcpy
(
label
,
&
labels_
[
currentSampleIndex_
],
sizeof
(
int
)
*
n
);
currentSampleIndex_
+=
n
;
...
...
paddle/gserver/dataproviders/DataProvider.h
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/dataproviders/DataProviderGroup.h
浏览文件 @
80c68d38
...
...
@@ -12,7 +12,6 @@ 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. */
#pragma once
#include "DataProvider.h"
...
...
@@ -65,8 +64,8 @@ void DataProviderGroup<T>::reset() {
provider_
=
nullptr
;
// shuffle file list
std
::
shuffle
(
fileList_
.
begin
(),
fileList_
.
end
(),
ThreadLocalRandomEngine
::
get
());
std
::
shuffle
(
fileList_
.
begin
(),
fileList_
.
end
(),
ThreadLocalRandomEngine
::
get
());
startLoader
();
DataProvider
::
reset
();
...
...
@@ -113,8 +112,9 @@ void DataProviderGroup<T>::startLoader() {
size_t
endPos
=
std
::
min
(
fileList_
.
size
(),
startPos
+
loadFileCount
);
std
::
vector
<
std
::
string
>
fileVec
(
fileList_
.
begin
()
+
startPos
,
fileList_
.
begin
()
+
endPos
);
loader_
->
addJob
([
this
,
fileVec
]()
->
ProviderPtrType
{
return
this
->
loadFile
(
fileVec
);
});
loader_
->
addJob
([
this
,
fileVec
]()
->
ProviderPtrType
{
return
this
->
loadFile
(
fileVec
);
});
}
loader_
->
stopAddJob
();
}
...
...
paddle/gserver/dataproviders/MultiDataProvider.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/dataproviders/MultiDataProvider.h
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/dataproviders/ProtoDataProvider.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/dataproviders/ProtoDataProvider.h
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/dataproviders/ProtoReader.h
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/dataproviders/PyDataProvider.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/dataproviders/PyDataProvider.h
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/dataproviders/PyDataProvider2.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/evaluators/CTCErrorEvaluator.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/evaluators/ChunkEvaluator.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/evaluators/Evaluator.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/evaluators/Evaluator.h
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/gradientmachines/GradientMachine.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/gradientmachines/GradientMachine.h
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/gradientmachines/GradientMachineMode.h
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/gradientmachines/MultiGradientMachine.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/gradientmachines/MultiGradientMachine.h
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/gradientmachines/MultiNetwork.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/gradientmachines/MultiNetwork.h
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/gradientmachines/NeuralNetwork.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/gradientmachines/NeuralNetwork.h
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/gradientmachines/ParallelNeuralNetwork.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/gradientmachines/ParallelNeuralNetwork.h
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/gradientmachines/RecurrentGradientMachine.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/gradientmachines/RecurrentGradientMachine.h
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/layers/AddtoLayer.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/layers/AddtoLayer.h
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/layers/AgentLayer.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/layers/AgentLayer.h
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/layers/AverageLayer.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/layers/BatchNormBaseLayer.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/layers/BatchNormBaseLayer.h
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/layers/BatchNormalizationLayer.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/layers/BatchNormalizationLayer.h
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/layers/BilinearInterpLayer.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/layers/BlockExpandLayer.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/layers/BlockExpandLayer.h
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/layers/CRFDecodingLayer.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/layers/CRFDecodingLayer.h
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/layers/CRFLayer.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/layers/CRFLayer.h
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/layers/CTCLayer.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/layers/CTCLayer.h
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/layers/ConcatenateLayer.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/layers/ContextProjection.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/layers/ContextProjection.h
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/layers/ConvBaseLayer.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/layers/ConvBaseLayer.h
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/layers/ConvOperator.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/layers/ConvProjection.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/layers/ConvProjection.h
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/layers/ConvShiftLayer.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/layers/ConvexCombinationLayer.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/layers/CosSimLayer.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/layers/CosSimLayer.h
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/layers/CosSimVecMatLayer.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/layers/CostLayer.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/layers/CostLayer.h
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/layers/CudnnBatchNormLayer.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/layers/CudnnBatchNormLayer.h
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/layers/CudnnConvLayer.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/layers/CudnnConvLayer.h
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/layers/CudnnPoolLayer.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/layers/CudnnPoolLayer.h
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/layers/DataLayer.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/layers/DataLayer.h
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/layers/DataNormLayer.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/layers/DataNormLayer.h
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/layers/DotMulOperator.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/layers/DotMulProjection.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/layers/EosIdCheckLayer.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/layers/ExpandConvBaseLayer.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/layers/ExpandConvBaseLayer.h
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/layers/ExpandConvLayer.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/layers/ExpandConvLayer.h
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/layers/ExpandConvTransLayer.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/layers/ExpandConvTransLayer.h
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/layers/FeatureMapExpandLayer.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/layers/FullMatrixProjection.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/layers/FullMatrixProjection.h
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/layers/FullyConnectedLayer.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/layers/FullyConnectedLayer.h
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/layers/GatedRecurrentLayer.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/layers/GatedRecurrentLayer.h
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/layers/GetOutputLayer.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/layers/GruCompute.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/layers/GruCompute.h
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/layers/GruStepLayer.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/layers/HierarchicalSigmoidLayer.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/layers/HierarchicalSigmoidLayer.h
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/layers/IdentityProjection.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/layers/InterpolationLayer.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/layers/Layer.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/layers/Layer.h
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/layers/LinearChainCRF.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/layers/LinearChainCRF.h
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/layers/LinearChainCTC.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/layers/LinearChainCTC.h
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/layers/LstmCompute.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/layers/LstmCompute.h
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/layers/LstmLayer.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/layers/LstmLayer.h
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/layers/LstmStepLayer.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/layers/MDLstmLayer.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/layers/MaxIdLayer.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/layers/MaxLayer.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/layers/MaxLayer.h
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/layers/MixedLayer.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/layers/MixedLayer.h
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/layers/MultinomialSampler.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/layers/MultinomialSampler.h
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/layers/MultiplexLayer.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/layers/NCELayer.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/layers/NormLayer.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/layers/NormLayer.h
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/layers/NormProjectionLayer.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/layers/NormProjectionLayer.h
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/layers/Operator.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/layers/Operator.h
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/layers/OuterProdLayer.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/layers/ParameterReluLayer.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/layers/ParameterReluLayer.h
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/layers/PoolLayer.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/layers/PoolLayer.h
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/layers/PoolProjection.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/layers/PoolProjection.h
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/layers/PoolProjectionLayer.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/layers/PowerLayer.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/layers/PrintLayer.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/layers/Projection.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/layers/Projection.h
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/layers/RecurrentLayer.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/layers/RecurrentLayerGroup.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/layers/ResizeLayer.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/layers/ScalingLayer.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/layers/ScalingProjection.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/layers/SelectiveFullyConnectedLayer.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/layers/SelectiveFullyConnectedLayer.h
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/layers/SequenceConcatLayer.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/layers/SequenceLastInstanceLayer.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/layers/SequencePoolLayer.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/layers/SequenceReshapeLayer.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/layers/SequenceToBatch.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/layers/SequenceToBatch.h
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/layers/SlopeInterceptLayer.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/layers/SpatialPyramidPoolLayer.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/layers/SpatialPyramidPoolLayer.h
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/layers/SubSequenceLayer.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/layers/SumToOneNormLayer.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/layers/TableProjection.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/layers/TableProjection.h
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/layers/TensorLayer.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/layers/TensorLayer.h
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/layers/TransLayer.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/layers/TransLayer.h
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/layers/TransposedFullMatrixProjection.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/layers/ValidationLayer.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/tests/LayerGradUtil.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/tests/LayerGradUtil.h
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/tests/TestUtil.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/tests/TestUtil.h
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/tests/test_ActivationGrad.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/tests/test_ConvTrans.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/tests/test_Evaluator.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/tests/test_LayerGrad.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/tests/test_LinearChainCRF.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/tests/test_MultinomialSampler.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/tests/test_NetworkCompare.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/tests/test_ProtoDataProvider.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/tests/test_PyDataProvider.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/tests/test_PyDataProvider2.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/tests/test_RecurrentGradientMachine.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/tests/test_RecurrentLayer.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/gserver/tests/test_SelectiveFCLayer.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/math/Allocator.h
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/math/BaseMatrix.h
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/math/CpuSparseMatrix.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/math/CpuSparseMatrix.h
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/math/ExecViaCpu.h
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/math/MathFunctions.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/math/MathFunctions.h
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/math/MathUtils.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/math/MathUtils.h
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/math/Matrix.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/math/Matrix.h
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/math/MatrixBitCode.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/math/MemoryHandle.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/math/MemoryHandle.h
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/math/PoolAllocator.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/math/PoolAllocator.h
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/math/SIMDFunctions.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/math/SIMDFunctions.h
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/math/SparseMatrix.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/math/SparseMatrix.h
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/math/SparseRowMatrix.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/math/SparseRowMatrix.h
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/math/Storage.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/math/Vector.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/math/Vector.h
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/math/tests/test_Allocator.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/math/tests/test_ExecViaCpu.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/math/tests/test_FPException.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/math/tests/test_SIMDFunctions.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/math/tests/test_batchTranspose.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/math/tests/test_matrix.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/math/tests/test_matrixCompare.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/math/tests/test_matrixUtil.h
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/math/tests/test_perturbation.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/math/tests/test_sparseMatrixCompare.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/parameter/Argument.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/parameter/Argument.h
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/parameter/AverageOptimizer.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/parameter/AverageOptimizer.h
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/parameter/FirstOrderOptimizer.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/parameter/FirstOrderOptimizer.h
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/parameter/LearningRateScheduler.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/parameter/LearningRateScheduler.h
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/parameter/OptimizerFunctions.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/parameter/OptimizerFunctions.h
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/parameter/OptimizerWithRegularizer.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/parameter/OptimizerWithRegularizer.h
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/parameter/ParallelParameter.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/parameter/ParallelParameter.h
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/parameter/Parameter.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/parameter/Parameter.h
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/parameter/ParameterOptimizer.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/parameter/ParameterOptimizer.h
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/parameter/ParameterUpdateFunctions.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/parameter/ParameterUpdateFunctions.h
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/parameter/ParameterUpdaterBase.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/parameter/ParameterUpdaterBase.h
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/parameter/ParameterUpdaterHook.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/parameter/ParameterUpdaterHook.h
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/parameter/Regularizer.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/parameter/Regularizer.h
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/parameter/Weight.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/parameter/tests/test_common.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/pserver/BaseClient.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/pserver/BaseClient.h
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/pserver/LightNetwork.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/pserver/LightNetwork.h
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/pserver/ParameterClient2.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/pserver/ParameterClient2.h
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/pserver/ParameterServer2.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/pserver/ParameterServer2.h
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/pserver/ProtoServer.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/pserver/ProtoServer.h
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/pserver/RDMANetwork.h
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/pserver/SocketChannel.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/pserver/SocketChannel.h
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/pserver/SparseParameterDistribution.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/pserver/test/SocketTest.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/pserver/test/test_ParameterServer2.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/pserver/test/test_ProtoServer.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/trainer/ParamUtil.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/trainer/ParamUtil.h
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/trainer/ParameterUpdater.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/trainer/ParameterUpdater.h
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/trainer/RemoteParameterUpdater.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/trainer/RemoteParameterUpdater.h
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/trainer/Tester.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/trainer/Tester.h
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/trainer/TesterConfig.h
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/trainer/ThreadParameterUpdater.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/trainer/ThreadParameterUpdater.h
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/trainer/Trainer.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/trainer/Trainer.h
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/trainer/TrainerConfigHelper.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/trainer/TrainerConfigHelper.h
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/trainer/TrainerInternal.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/trainer/TrainerInternal.h
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/trainer/TrainerInternalConfig.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/trainer/TrainerInternalConfig.h
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/trainer/TrainerMain.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/trainer/tests/picojson.h
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/trainer/tests/test_Compare.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/trainer/tests/test_CompareSparse.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/trainer/tests/test_CompareTwoNets.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/trainer/tests/test_CompareTwoOpts.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/trainer/tests/test_Prediction.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/trainer/tests/test_PyDataProviderWrapper.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/trainer/tests/test_Trainer.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/trainer/tests/test_TrainerOnePass.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/trainer/tests/test_recurrent_machine_generation.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/utils/BarrierStat.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/utils/BarrierStat.h
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/utils/ClassRegistrar.h
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/utils/CommandLineParser.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/utils/CommandLineParser.h
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/utils/CustomStackTrace.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/utils/CustomStackTrace.h
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/utils/DisableCopy.h
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/utils/Excepts.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/utils/Flags.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/utils/Flags.h
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/utils/GlobalConstants.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/utils/GlobalConstants.h
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/utils/Locks.h
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/utils/Logging.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/utils/Logging.h
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/utils/PythonUtil.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/utils/PythonUtil.h
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/utils/Queue.h
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/utils/Stat.h
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/utils/StringUtil.h
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/utils/Thread.h
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/utils/ThreadLocal.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/utils/ThreadLocal.h
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/utils/TypeDefs.h
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/utils/Util.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/utils/Util.h
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/utils/Version.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/utils/Version.h
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/utils/arch/linux/Locks.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/utils/arch/osx/Locks.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/utils/tests/test_CommandLineParser.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/utils/tests/test_CustomStackTrace.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/utils/tests/test_CustomStackTracePrint.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/utils/tests/test_Logging.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/utils/tests/test_SpinLock.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/utils/tests/test_StringUtils.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/utils/tests/test_Thread.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
paddle/utils/tests/test_ThreadBarrier.cpp
浏览文件 @
80c68d38
此差异已折叠。
点击以展开。
编辑
预览
Markdown
is supported
0%
请重试
或
添加新附件
.
添加附件
取消
You are about to add
0
people
to the discussion. Proceed with caution.
先完成此消息的编辑!
取消
想要评论请
注册
或
登录