Jerry Wu
36685c3226
Caffe persister ( #1095 )
...
* add dynamic caffe loading support
* refined with bug fix and more commnets
* fix typo and bias check for convolution
* renaming change
* fix import reference
* refine customized layer support
* fix style check issue
* resolve style issue
* resolve conflict
* caffe persister implementation
* refinement with pr
* rename loadcaffe name
* fix typo
* add docs
2017-07-11 14:35:31 +08:00
Xianyan
dfc5958854
[Issue 864] save module weights, load (all / part) module weights ( #1102 )
...
* load save module weights
* add support for loading from another model
* meet code review
2017-07-11 13:57:59 +08:00
JerryYanWan
f4170c89ef
fix bugs in convLSTM
2017-07-11 12:56:53 +08:00
ding
b15db4690b
fix comments
2017-07-07 09:47:14 -07:00
ding
2b9d72caea
support convlstm and convolution without bias
2017-07-07 09:47:14 -07:00
ding
ea6ad9dd90
fix comments
2017-07-06 19:38:01 -07:00
ding
7bb0d92a68
support spatial convultion without bias
2017-07-06 19:38:01 -07:00
Yao Zhang
6d7f846894
Remove unnecessary varialbe in Recurrent object construtor ( #1127 )
...
* add test
* remove unnecessary change in LinearSpec
* Remove unnecessary varialbe in Recurrent object construtor
* pass the test
2017-07-07 10:12:11 +08:00
Yao Zhang
9f00c87f7d
Make SpatialFullConvolution declare with implict numeric type ( #1128 )
...
* add test
* remove unnecessary change in LinearSpec
* Make SpatialFullConvolution declare with implict numeric type
* rebase master
* change SpatailaFullConvolution doc
2017-07-06 18:47:07 +08:00
Yan Wan
2f25b70660
Optimize LSTMPeephole by pre-calculating the input2hidden linear ( #1158 )
...
* optimize LSTMPeephole
* remove setname
* test speed, and add class variable
* rename timeDim to featDim in LSTMPeephole
2017-07-06 16:55:12 +08:00
Yan Wan
ffae7fa1ba
optimize GRU by pre-calculating three linear operations ( #1150 )
...
* add preTopology for GRU
optimize GRU
* remove revisions in peephole
* pass dropout test
* add speed test
* remove unnecessary variables; judge preTopology == Sequential
* add comment
2017-07-06 15:37:28 +08:00
Yan Wan
5e852b278c
Timedistributed add getParametersTable method ( #1154 )
...
* override getParametersTable method
* test getParametersTable
2017-07-05 16:21:50 +08:00
JerryYanWan
ed95a2be0f
revise hashCode to avoid null pointer
2017-07-05 09:49:45 +08:00
Ian Wong
4f92f58e2e
Checkin document pages ( #1068 )
...
* checkin md doc files
* add some docs
* add more docs
* change the folder location
* fix unit test
* fix typos
* fix typo
2017-07-04 13:52:13 +08:00
JerryYanWan
44d65c765b
fix set bugs for fInput and fGradInput in TimeDistributed layer
...
code style
2017-07-03 17:34:47 +08:00
Xianyan
9d9ea0c2f4
fix VolumetricConvolution unit test, and fix bug in batch mode ( #1105 )
...
* fix VolumetricConvolution unit test, and fix bug in batch mode
* meet pr review
2017-06-30 15:41:47 +08:00
Xin Qiu
6c09fdc644
fix some unit test ( #1104 )
2017-06-30 13:46:11 +08:00
zhangxiaoli73
0fea118680
makedown docs for some API ( #1075 )
...
* add some api md
* change according to reviews
2017-06-30 13:37:50 +08:00
yangw
a75550b1da
fix RnnCell factory method
2017-06-29 05:10:05 -05:00
Yan Wan
96528b334f
add preTopology to boost performance of rnn ( #1117 )
...
* add preTopology to boost performance of rnn
* require only one Cell added
* require logic revision, modules.length == 1 || modules.length == 2
* add Javadoc for preTopology in Cell and Recurrent
2017-06-29 16:02:47 +08:00
xinyue
ae0ba8a35d
[Issue 930] Separate scale into scaleW and scaleB ( #975 )
2017-06-27 13:13:10 +08:00
Yao Zhang
20517a36c7
Polish InferReshape documentation ( #1087 )
...
* Polish InferReshape documentation
* remove linearSpec unnecessary change
* fix core conflict
* meet code review
2017-06-26 14:07:58 +08:00
zhangxiaoli73
df08717f07
add L1HingeEmbeddingCriterion unit test for python ( #1073 )
2017-06-23 16:52:47 +08:00
Ian Wong
ca2dce6049
rename apply to inputs when define connection in Graph API ( #1069 )
...
* rename apply to inputs
* fix style check
2017-06-23 13:40:19 +08:00
Xianyan
ff2ada5f87
let getNamedModule also return containers, add unit test for it ( #1067 )
2017-06-23 10:17:16 +08:00
Jerry Wu
1210ae50fe
Caffe Model Support - Load Caffe Dynamically from Prototxt and Binary ( #964 )
...
* add dynamic caffe loading support
* refined with bug fix and more commnets
* fix typo and bias check for convolution
* renaming change
* fix import reference
* refine customized layer support
* fix style check issue
* resolve style issue
2017-06-23 10:10:57 +08:00
zhangxiaoli73
9a643ab0c8
change Table to Tensor in L1HingeEmbeddingCriterion ( #1061 )
2017-06-22 16:33:53 +08:00
Xianyan
0730824d61
[Issue 1035] Allow layer regulazier set after model definition, add regulazier for cadd and cmul ( #1048 )
...
* Allow layer regulazier set after model definition
* add regulazier support for cadd and cmul
* meet code review
* add python set API, add regularizer to python cmul and cadd
* add test for setBRegularizer
* add parameters to create method, add doc
2017-06-22 11:50:07 +08:00
Yuchao-Tao
f695785ab0
Python regularizer ( #1008 )
...
* # This is a combination of 5 commits.
# The first commit's message is:
add Regularizer to PythonBidgDL.scala and optimizer.py
# This is the 2nd commit message:
update regularizer in layer.py
# This is the 3rd commit message:
some fix
# This is the 4th commit message:
some fix 2
# This is the 5th commit message:
adjust according to the test
* Resolve Conflicts; fix; fix RNN; fix; pass compilation; fix; fix on the missmatch on the parameter list
* fix style
* fix style
* fix style
* fix style
2017-06-22 09:24:02 +08:00
Xin Qiu
1525c950d0
update some doc ( #1050 )
2017-06-21 13:22:12 +08:00
Nisha Muktewar
bf25d2de72
Patch to improve error message when the target (label) value starts at 0 ( #1049 )
...
* Changes to SpatialConvolution.scala to avoid divide by 0 error
* reversing changes
* Update error mesg when target value starts at 0
2017-06-21 13:21:24 +08:00
Ian Wong
016b668171
Support Tensorflow model file read/write ( #800 )
...
* nn refactor
* nn refactor
* fix code style issue
* change back the layers
* code refactor
* change tests to automatic test
* add more test for model save
* remove some useless unit test
* add more save test
* add more writer test
* rnn test case automation
* refine save test
* refine save unit test
* remove NHWC
* meet code review
* meet code review
* use MulConst in MulTF
* add a flatten node for tf 1.1
* fix code style and failed unit test
* mv tf model layers to another package
* add a python example to show how to define model in tensorflow and run in BigDL
* move tf example python code to example folder
* Add backward test in LeNet and AlexNet (#19 )
* add unit test of lenet backward
* add some print
* add backward test in lenet and alexnet
* seperate testModel into forward and backward methods
2017-06-21 12:50:17 +08:00
Yan Wan
9e94774514
using set instead of copy for output and gradInput ( #1028 )
...
* using set instead of copy for output and gradInput
define private method
* revise set method
* add condition before set method
* grammar revision
* set cellAppendStartIdx to cells.length
* GRU, last layer CAddTable(false)
2017-06-20 14:41:51 +08:00
li,zhichao
5cf02b6bc0
Fix HardShrink factory method ( #1025 )
...
* fix HardShrink factory
* update unittest
2017-06-16 11:26:39 +08:00
qiuxin2012
98535688fe
fix concat return incorrect result when output is 1D
2017-06-15 16:43:46 +08:00
zhangxiaoli73
c67f7f4a76
unchange Euclidean input ( #991 )
2017-06-09 16:37:50 +08:00
yangw
54c1021b0d
add python api
2017-06-08 22:39:22 +08:00
Xianyan
c58a6ab420
refactor smoothL1Criterion and fix tensor storage usage ( #986 )
...
* refactor smoothL1Criterion and fix tensor storage usage
2017-06-08 12:36:15 +08:00
zhangxiaoli73
970763f895
delete useless CriterionTable ( #981 )
2017-06-08 09:51:02 +08:00
Ian Wong
17349e48d9
some new layer and existing layer refactor ( #957 )
...
* nn refactor
* fix code style issue
* add python api
* change back the layers
2017-06-07 19:10:34 +08:00
Yang Wang
f6afb611b0
refactor InitializationMethod ( #960 )
...
* refactor init
* fix test
* address some comments
* fix tests
* add trait
* fix tests
2017-06-07 18:08:12 +08:00
Fu Zhouwang
1342865e4d
Add isTensor and isTable method in Activity with additional test cases ( #980 )
...
* Add isTensor and isTable method in Activity
* fix sytle in comment
2017-06-07 15:08:20 +08:00
Xianyan
8437b7b12c
change SelectTable from Container to AbstractModule ( #977 )
2017-06-07 09:44:31 +08:00
zhangxiaoli73
65afa653be
fix add bug ( #968 )
2017-06-06 15:34:20 +08:00
Yan Wan
7b9e5cbd15
share memory in CAddTable for LSTM, GRU, SimpleRNN ( #855 )
...
* share memory in CAddTable for LSTM, GRU, SimpleRNN; share memory for Reverse layer
* revise javadoc
* revert GRU
* set default inPlace value to be false in Reverse layer
2017-06-06 14:52:08 +08:00
qiuxin2012
776009b736
fix BCECriterion
2017-06-05 16:06:24 +08:00
jenniew
9c39513289
add tensor size info when requirement failed ( #942 )
...
* add tensor size info when requirement failed
* add layer info when requirement fail
2017-06-01 06:37:49 +08:00
Yao Zhang
fc15ce4521
fix bug in Concat layer and Add backward to ConcatTable ( #902 )
2017-05-25 10:48:32 +08:00
qiuxin2012
00d473811a
fix ConcatTable.clearState() don't clear gradInput when gradInput is a table
2017-05-25 10:42:35 +08:00
JerryYanWan
52cc3d2fbe
linear; initialize gradWeight when running backward
2017-05-24 12:57:55 +08:00
