a8526f105d
Merge remote-tracking branch 'upstream/develop' into factorization_machine_layer
8 years ago
4c72b0634c
Fix creation of tmp variable in factorization machine layer
8 years ago
46f7854e5c
fix mac compile ( #4870 )
8 years ago
a30d53b792
Merge remote-tracking branch 'upstream/develop' into factorization_machine_layer
8 years ago
60b84856e7
refine the conf files, combine in one file
8 years ago
9741ade8ee
Change pow to square in factorization machine layer
8 years ago
b487c62aac
Merge remote-tracking branch 'upstream/develop' into merge_grad_gtest
8 years ago
2dc6cea65c
add unit tests for mkldnn branches, and fix typo
8 years ago
86053e7766
Reduce the input size in testing factorization machine
8 years ago
2ce8f1875b
Fix tests for factorization machine layer
8 years ago
22c5d1f147
Merge remote-tracking branch 'upstream/develop' into factorization_machine_layer
8 years ago
947b6a77ce
Implement factorization machine layer
8 years ago
1199aa6876
fix bug: clear grad and always share data when output has cpu
...
and add activation in unit tests
8 years ago
dee7f81325
add finish work of mkldnn
8 years ago
60c2536d22
Merge branch 'develop' of https://github.com/PaddlePaddle/Paddle into develop
8 years ago
7a7c8fd9e5
simplify some comments and code
8 years ago
e1954857d7
fix bug: merge grad must before backward act.
...
and add branch net comparing with cpu result
8 years ago
7f0c823093
lambda cost bug's fix
8 years ago
698071cc33
share mkldnn output value data if next layer is cpu device
8 years ago
6715beaace
enable merge output grad of mkldnn
8 years ago
f504c8a83d
Remove unnecessary configs
8 years ago
5e78c7aea6
Merge remote-tracking branch 'upstream/develop' into factorization_machine_layer
8 years ago
1644c72acc
Add framework of the factorization machine layer
8 years ago
aa1e92ec9a
Merge branch 'develop' of https://github.com/baidu/Paddle into inference
8 years ago
8f2c48cf4c
Fix Layer.cpp
8 years ago
adad8d9ed2
Open WITH_TESTING option.
8 years ago
4558807c48
Use PADDLE_WITH_CUDA instead of PADDLE_WITH_GPU
8 years ago
84500f9487
Change `PADDLE_ONLY_CPU` to `PADDLE_WITH_GPU`
...
By shell command
```bash
sed -i 's#ifdef PADDLE_ONLY_CPU#ifndef PADDLE_WITH_GPU#g' `find ./paddle/ -name '*.h' -o -name '*.cc' -o -name '*.cpp' -o -name '*.c' -o -name '*.cu'`
sed -i 's#ifndef PADDLE_ONLY_CPU#ifdef PADDLE_WITH_GPU#g' `find ./paddle/ -name '*.h' -o -name '*.cc' -o -name '*.cpp' -o -name '*.c' -o -name '*.cu'`
```
8 years ago
bb07120b64
Remove dataproviders.
8 years ago
3c66b307f7
Remove the pserver, trainer, evaluators and some useless gradientmachines when compile mobile inference library.
8 years ago
0a37f2adf0
fix typo
8 years ago
c53d214205
fix compiler warning from MKLDNNLayer and so on
8 years ago
672c968176
remove gpu code when backward mkldnn_softmax
8 years ago
2c6ac6297d
refine code and remove min clip
8 years ago
7483087c8c
enable mkldnn_softmax
8 years ago
bea39f6314
add compare simple net of mkldnn and cpu
8 years ago
ed27a3be78
add some log info
8 years ago
4400284685
Merge pull request #4224 from tensor-tang/act
...
some addition for mkldnn_act
8 years ago
4ec3a77a43
should run resetBwd before bwdAct
8 years ago
fa722385de
refine test_MKLDNN and skip memory copy for relu
8 years ago
8deda323cd
Merge pull request #4197 from luotao1/refine_layer
...
remove unused XXXLayer.h in Layer.cpp
8 years ago
69c613fb5a
refine Layer.cpp for some CostLayer
8 years ago
a9202e89cf
Merge pull request #3764 from Xreki/build_ios
...
Add the support of the cross-compiling for iOS devices
8 years ago
c15ac20257
Merge pull request #4199 from tensor-tang/mkldnn_act
...
add MKLDNN relu and tanh
8 years ago
9d692e3bcc
add gtest for MKLDNN activation and pass them
8 years ago
70141a73c4
remove unused XXXLayer.h in Layer.cpp
8 years ago
24f13b1a59
add MKLDNN activation files
8 years ago
17bc593ca2
Update DetectionOutputLayer.cpp
...
Add a space
8 years ago
d3deb6faba
Update DetectionOutputLayer.cpp
...
Fix format
8 years ago
71b3fbb18a
Fix a ssd bug
8 years ago
49d4b39f28
fix typo and remove some unused code
8 years ago
37268dbd33
update using exclusive padding for avg-pool
8 years ago
cc28fb4bb3
Merge remote-tracking branch 'upstream/develop' into mkldnn_pool
8 years ago
cc7a7557a2
set CudnnPooling don't include padded values in default
8 years ago
b012013a58
Revert "Do not invoke GPU method when use_gpu=false"
8 years ago
9d108a21a7
add unit test for mkldnn_pool and pass them
8 years ago
17e1aa8d43
Do not invoke GPU method when use_gpu=false
8 years ago
2a98cba2d2
enable mkldnn_pool forward and backward
8 years ago
b1bca06667
Refine the ExpandConvLayer.
8 years ago
e4c340e4e7
Bug fix.
8 years ago
88b31422b3
Delete ExpandConvBaseLayer.cpp and ExpandConvBaseLayer.h
8 years ago
3616ef7739
Remove ExpandConvBaseLayer.
8 years ago
b3a50d53c9
Remove useless code.
8 years ago
2a110af309
Use Matrix::addBias.
8 years ago
d3a86c673b
init mkldnn_pool files
8 years ago
b0d9b68a5f
unify functions of mkldnn_fc and refine comments
8 years ago
654344b94c
Merge pull request #4032 from tensor-tang/mkldnn-conv
...
Add MKLDNNConvLayer
8 years ago
f2317b67f7
separate resetFwd and resetBwd to some sub functions
8 years ago
9e524fce1b
Remove the generation of executable, such as paddle_trainer, when setting WITH_C_API.
8 years ago
66fdbd0cee
add some comment and simplify some code
8 years ago
e76eb8534d
fix SwitchOrderLayer forward
8 years ago
c39b771a53
add test cases for mkldnn_conv
8 years ago
f3a23b6840
add MKLDNNConvLayer
8 years ago
171fee2c9e
Merge pull request #4011 from tensor-tang/refine
...
Refine MKLDNNLayer and MKLDNNTester
8 years ago
7f7fa32528
use CPU_DEVICE instead of magic number
8 years ago
4f0869bef3
add some necessary params for reset functions
8 years ago
d131152b28
Merge pull request #4021 from chengduoZH/fix_DeConv3D_switch_imgSize_and_output_
...
fix DeConv3D switch(imgSize*_, output*_)
8 years ago
e9a1f2ad8b
fix DeConv3D switch(imgSize*_, output*_)
8 years ago
f3bb7b99dd
refine MKLDNNTester add UpdateCallback for test
8 years ago
94ea8ee0e5
refine MKLDNNLayer logical: move forward and backward to MKLDNNLayer and remove copyOutputInfoToOtherDevice
8 years ago
f40d5f580d
remove syncOutputGrad, rename syncInputValue to updateInputData
8 years ago
d4c0734840
remove convertOutputToOtherDevice
8 years ago
d74fe78040
refine MKLDNNMatrix, solid data handle, rename updateData to setData
8 years ago
38e7ad799c
Set detection output to NULL
8 years ago
4343be3309
Update DetectionUtil.cpp
8 years ago
dade4272c7
Update DetectionUtil.cpp
8 years ago
a510377024
Fix ssd bugs.
8 years ago
0be349496f
Merge pull request #3594 from chengduoZH/Adapting_to_the_BatchNorm_structure_to_support_3D_data
...
Adapting to the BatchNorm structure to support 3D data
8 years ago
91a0c11b19
Adaptive data structure for SwitchOrderLayer
8 years ago
fa54d9d8c0
Rename attribute name in proto
8 years ago
bb56b58376
Fix warning log in switch order layer
8 years ago
0146e8bfc5
Merge pull request #3843 from Xreki/build_android_clang
...
Enable the use of clang to cross-compile for Android
8 years ago
de145b0aa5
Merge branch 'develop' of https://github.com/PaddlePaddle/Paddle into pixel_softmax_layer
8 years ago
4373a4a74d
fix redeclaration of 'paddle::StatPtr __stat'
8 years ago
5500153a6d
fix cudnnBatchNorm for 3D data
8 years ago
3d490d030e
Merge branch 'develop' of https://github.com/PaddlePaddle/Paddle into pixel_softmax_layer
8 years ago
f2f839af27
Merge pull request #3874 from chengduoZH/fix_Conv3d_DeConv3d
...
Fix conv3d DeConv3d (bias shape)
8 years ago
cdae0c754e
fix Conv3d, DeConv3d (bias shape)
8 years ago
a523bea8e5
fix getType.
8 years ago
a162ebce8b
Merge branch 'develop' into pixel_softmax_layer
8 years ago
0b478e991c
follow comments.
8 years ago
faa4da4835
fix ctc edit distance in v2 API.
8 years ago
2fc012c54c
Merge branch 'develop' into build_android_clang
8 years ago
fb69d38c4b
Merge branch 'develop' of https://github.com/PaddlePaddle/Paddle into Adapting_to_the_BatchNorm_structure_to_support_3D_data
8 years ago
ba1aa60700
fix the problem that long message will be truncated.
8 years ago
8b15ac82fa
Move the definition of hl_cpu_gru_forward and hl_cpu_gru_backward to function/GruFunctor.h.
8 years ago
92c293aca8
Merge pull request #3794 from Xreki/fix_mac_compiling_std_move
...
Update the version of OpenBLAS and fix compiling error caused by std::move on Mac
8 years ago
988e49defa
Fix clang compiling error of "moving a temporary object prevents copy elision" causing by std::move, and fix some warnings.
8 years ago
a55dd22649
fix conflict
8 years ago
6374831966
fix conflict
8 years ago
a357bd6fe5
Fix conflict and Add test_pooling3D_layer.protostr
8 years ago
455d2bd7b1
fix conflict
8 years ago
536dbc0cb6
fix conflict
8 years ago
36f0aa7390
fix code style to pass CI.
8 years ago
3d1b87193d
Merge branch 'develop' into cross_entropy_over_beam
8 years ago
09e903eb94
fix v2 infer interface.
8 years ago
0c273ff41e
fix conflict
8 years ago
2563e32bb1
fix clang build error
8 years ago
43da7675b3
fix conflict
8 years ago
322d9ad81b
Merge pull request #3712 from tensor-tang/merge
...
add MKLDNN_DEVICE
8 years ago
b45d020faf
Merge pull request #3718 from hedaoyuan/convolution
...
Depthwise Convolution Optimization
8 years ago
c5183caa04
rename
8 years ago
168707cadd
Fix a small bug.
8 years ago
3cf01b5d52
refine ROIPoolLayer
8 years ago
4dcc533206
fix conflict
8 years ago
bfbd066fdd
refine
8 years ago
b709af616f
HuberTwoClassification only support one dimension
8 years ago
6053f7e36b
fix previous comments(c++)
8 years ago
e63ad0a6bd
HuberRegressionLoss and HuberTwoClassification support multi-dimension data
8 years ago
34a92ab41a
ExpandConvLayer adds support of arm-neon acceleration.
8 years ago
acd8a22ed1
Merge branch 'develop' into huber_loss
8 years ago
2efac83aaa
Merge remote-tracking branch 'upstream/develop' into merge
8 years ago
98b7c67364
add todo
8 years ago
3d12a610ca
Merge branch 'develop' into cross_entropy_over_beam
8 years ago
4cc57836f3
enable reorder
8 years ago
14194ba08d
Merge pull request #3679 from lcy-seso/fix_bug_of_seq_slice_layer
...
fix a bug of sequence_slice layer when batch_size=1.
8 years ago
aa28d046fb
fix a bug of sequence_slice layer when batch_size=1
8 years ago
7a42c92d49
fix a bug that memory does not clean.
8 years ago
db39935b77
Merge pull request #3647 from yu239/fill_gen_output
...
fix ids reverse order in fillGenOutputs.
8 years ago
860bf192c9
Add maxPoolIdx
8 years ago
7596191821
Merge branch 'develop' into huber_loss
8 years ago
790379f147
fix above comments
8 years ago
e286b950a5
Merge branch 'develop' of https://github.com/PaddlePaddle/Paddle into dilate_conv
8 years ago
11f4283f09
Merge pull request #3634 from reyoung/feature/minus_op
...
Feature/minus op
8 years ago
7fdf5c831e
Merge pull request #3645 from jacquesqiao/reduce-warning
...
reduce some compile warning
8 years ago
dee4c832cc
Merge branch 'develop' of github.com:baidu/Paddle into feature/minus_op
8 years ago
74d3ca8bab
Merge branch 'develop' into cross_entropy_over_beam
8 years ago
ab6b3c481a
Merge pull request #3367 from lcy-seso/add_sequence_slice_layer
...
Add a sequence slice layer.
8 years ago
3b6069b7f5
fix ids reverse order in fillGenOutputs
8 years ago
72d29186bb
reduce some compile warning
8 years ago
5e59ca7ccc
fix config helper.
8 years ago
057694aac9
Merge branch 'develop' of https://github.com/PaddlePaddle/Paddle into dilate_conv
8 years ago
e3342ff8e7
Fix android build error.
8 years ago
780c8d969e
make downSpatial work, and remove hasSpatial_
8 years ago
48d87e5e91
pass test, support input CPU device
8 years ago
377401fb0c
fix a bug.
8 years ago
bc6bae8d53
Merge branch 'develop' into add_sequence_slice_layer
8 years ago
82e4fab4e3
follow comments.
8 years ago
2c8b3c4430
Merge branch 'develop' of https://github.com/PaddlePaddle/Paddle into dilate_conv
8 years ago
d8cd67dd1e
Make cudnn convolution layer and projection support for dilation.
8 years ago
3bf440023a
follow comments.
8 years ago
fc2e6f1ca0
Merge branch 'develop' into print_attention_weight
8 years ago
25083de998
Merge branch 'develop' into cross_entropy_over_beam
8 years ago
8f4ca2d12f
add implementations.
8 years ago
4eecd0c2d5
use MKLDNNMatrix in fc backward
8 years ago
6eab5638f0
Fix remote large update core ( #3518 )
...
* fix remote large update core
* wip
* working version
* fix style check
* fix style check
* update style check
8 years ago
93539093f4
Allow boot_bias for recurrent group to be static
8 years ago
2377d71947
Add3DPooling
8 years ago
73ab2d4678
fix backward error of huber_regression_cost
8 years ago
4bffbd30f0
use MKLDNNMatrix in fc forward
8 years ago
5d98b6f217
Adapting to the BatchNorm structure to support 3D data
8 years ago
d5768ebc89
fix above comments
8 years ago
8e4dcf8b7f
Merge branch 'develop' of https://github.com/PaddlePaddle/paddle into add-ShiftLayer
8 years ago
0af1c4a9fe
Follow comments and refine annotations on ScaleShiftLayer
8 years ago
83abbce8eb
Follow comments and refine ScaleShiftLayer
8 years ago
29d8825caf
tune relative precision for unit test img_conv2 in test_NetworkCompare.cpp.
...
1. It's no problem with relative precision 1e-3 when testing several times in my local machine.
2. But the testing failed with 1e-2 in the TeamCity, and only one value's relative precision is over 1e-2. So tune it to 4e-2
8 years ago
1c0a1a076d
Merge branch 'develop' into huber_loss
8 years ago
3065cb2625
add huber_regression_cost
8 years ago
38cc5dadcc
modified bias shape of ConvLayer
8 years ago
62e6dac402
add MKLDNNMatrix files
8 years ago
c792ef7d5a
fix DeConv3D, Conv3D
8 years ago
424b325d08
add unit test DeConv3D, Conv3D, col2vol, vol2col
8 years ago
7d2ef02a99
Add ScaleShiftLayer
8 years ago
07d16e3e13
refine comments
8 years ago
27a99bfb14
Add base class for huber_regression_cost and huber_classification_cost
8 years ago
e08651f9b5
remove flag use_mkldnn_wgt
8 years ago
a037b099f7
finish unittest.
8 years ago
c07cbf7daf
Merge branch 'develop' of https://github.com/PaddlePaddle/paddle into add-ROIPooling
8 years ago
3321dd034b
Merge branch 'develop' into huber_loss
8 years ago
e6db484d15
make clear that current huber_cost is for two-classification
8 years ago
1e828dc1de
Merge branch 'develop' into cross_entropy_over_beam
8 years ago
05e8a26b4b
add unittest.
8 years ago
759a9d3ab5
follow comments.
8 years ago
9c448be51e
enable self-defined index data in testLayerGrad.
8 years ago
23cf0c61e0
Add DeConv3DLayer
8 years ago
11975b4f91
Add Conv3DLayer
8 years ago
5d7f6dde52
Add depth dimension information to ConvBaseLayer
8 years ago
b7359ee3ce
Merge branch 'develop' into add_sequence_slice_layer
8 years ago
936de12b85
Merge branch 'develop' into delete_useless_codes_in_softmax
8 years ago
b97f020f9c
fix unittest error.
8 years ago
cac4ad4449
delete useless codes in softmax backward.
8 years ago
2e87d747bd
Merge pull request #3337 from tensor-tang/merge
...
Enable mkldnn_fc for general format
8 years ago
2ca13944b6
Merge pull request #3378 from gangliao/proj_name
...
Rename PROJ_ROOT to PADDLE_SOURCE_DIR and PROJ_BINARY_ROOT to PADDLE_…
8 years ago
2d4c66d4b2
add comments and todo lists
8 years ago
a475a57d9b
rename files and classes, use uppercase of Mkldnn and Cpu
8 years ago
7a56d46a8a
Rename PROJ_ROOT to PADDLE_SOURCE_DIR and PROJ_BINARY_ROOT to PADDLE_BINARY_DIR
8 years ago
024243fee0
Fix typo error.
8 years ago
7ff689f55a
Merge branch 'develop' into add_sequence_slice_layer
8 years ago
b2bd67133a
rename and refine functions
8 years ago
26bc5b1259
add implementations.
8 years ago
32e756ca03
fix test_KmaxSeqScore for only CPU compile.
8 years ago
f6a940936b
remove unused comments, refine and rename
8 years ago
6c66044ce0
Fix more warnings about comparison between signed and unsigned values
8 years ago
4480115681
Fix some build warnings
8 years ago
abc49f74af
Merge remote-tracking branch 'upstream/develop' into merge
8 years ago
e18fbd8208
skip reset mkldnn when input size does not change
8 years ago
6373291c77
add test case use_mkldnn_wgt
8 years ago
34ff75221f
Merge branch 'develop' into add_sequence_slice_layer
8 years ago
2988a58ef0
add unittest.
8 years ago
92b2b1bd73
Merge branch 'develop' into kmax_score_layer
8 years ago
d9f97b023d
Merge pull request #3297 from lcy-seso/add_nest_sequence_select
...
Add a nest sequence select layer.
8 years ago
0c951176bd
pass mkldnn gtest
8 years ago
7da1db053b
update cuda kernel.
8 years ago
59a8ebc62a
Merge branch 'develop' into kmax_score_layer
8 years ago
3e4c671cd4
Merge branch 'develop' of https://github.com/PaddlePaddle/Paddle into bn_infer
8 years ago
bf08e5d985
modify code comments.
8 years ago
73192bb12a
add a batch norm inference kernel.
8 years ago
94686c5751
Merge branch 'develop' into add_nest_sequence_select
8 years ago
98a83cd280
follow comments.
8 years ago
42c102a0b3
follow comments.
8 years ago
ec9009f320
add mkldnn tester
8 years ago
ffafc5c911
fix the SubNestedSequenceLayer implementations.
8 years ago
3ce497ca5a
enhance cpplint cmake
8 years ago
1203ebc498
add mkldnn fc backward
8 years ago
90d5be7417
add mkldnn fc forward
8 years ago
94b172a7e8
fix mkldnn lib bug, and mkldnnbase
8 years ago
29fa73bc40
fix unittest.
8 years ago
455ad5b5b2
Merge pull request #3141 from hedaoyuan/nnpack
...
Support groups in NNPACKFunction.
8 years ago
fe04175ed3
Fix comparison between signed and unsigned values
8 years ago
2357aee368
Merge pull request #3250 from lcy-seso/seq_softmax_in_sub_seq_mode
...
add gradient check for sequence softmax activation.
8 years ago
d1e754336e
Merge branch 'clang-format' of https://github.com/gangliao/Paddle ; branch 'develop' of https://github.com/PaddlePaddle/Paddle into clang-format
8 years ago
89ea288c87
Merge branch 'develop' into kmax_score_layer
8 years ago
00b6d26684
Merge branch 'develop' into add_nest_sequence_select
8 years ago
927adb4a41
Merge branch 'develop' into kmax_score_layer
8 years ago
3c798828a9
SKip cpplint if source is not changed.
8 years ago
78a772ea64
Merge pull request #3234 from lcy-seso/define_subseq_data_in_gradient_check
...
enable defining sub-sequence data in test layer gradients.
8 years ago
855ae59d44
add KmaxSeqScoreLayer implementation.
8 years ago
3c3a11a0dc
add use_mkldnn flag
8 years ago
6bb970b52b
enable defining sub-sequence data in test layer gradients.
8 years ago
def959a890
add mkldnn fc files
8 years ago
wangmeng28
Qiao Longfei
tensor-tang
chengxingyi
hedaoyuan
Yi Wang
Yu Yang
Luo Tao
Yiqun Liu
Yuan Gao
gaoyuan
chengduoZH
chengduo
wenshilei
wanghaoshuang
hedaoyuan
qingqing01
caoying03
qijun
guosheng
Cao Ying
haonanyu
武毅
xuwei06
Yi Wang
liaogang
Yi Wang