[GitHub] asitstands opened a new pull request #9870: New example of custom operator using RTC

[GitBox]

asitstands opened a new pull request #9870: New example of custom operator 
using RTC
URL: https://github.com/apache/incubator-mxnet/pull/9870
 
 
   ## Description ##
   
   `example/numpy-ops/ndarray_softmax.py` contains a custom softmax 
implementation using RTC. However, it extends `NDArrayOp` which is deprecated. 
Moreover the example seems not compatible with the current RTC API. I think 
that the example confuses newbies like me who does not know the previous 
history of the API changes. So I think it is a good idea to remove the broken 
example and add a new RTC example based on `CustomOp`. The new example 
`example/numpy-ops/custom_softmax_rtc.py` is based on the existing example 
(`example/numpy-ops/custom_softmax.py`) but uses custom CUDA kernel through 
RTC. 
   
   ## Checklist ##
   ### Essentials ###
   - [ ] Passed code style checking (`make lint`)
   - [ ] Changes are complete (i.e. I finished coding on this PR)
   - [ ] All changes have test coverage:
   - Unit tests are added for small changes to verify correctness (e.g. adding 
a new operator)
   - Nightly tests are added for complicated/long-running ones (e.g. changing 
distributed kvstore)
   - Build tests will be added for build configuration changes (e.g. adding a 
new build option with NCCL)
   - [ ] Code is well-documented: 
   - For user-facing API changes, API doc string has been updated. 
   - For new C++ functions in header files, their functionalities and arguments 
are documented. 
   - For new examples, README.md is added to explain the what the example does, 
the source of the dataset, expected performance on test set and reference to 
the original paper if applicable
   - [ ] To the my best knowledge, examples are either not affected by this 
change, or have been fixed to be compatible with this change
   
   ### Changes ###
   - `example/numpy-ops/ndarray_softmax.py` is removed.
   - `example/numpy-ops/custom_softmax_rtc.py` is added.
   - `example/numpy-ops/README.md` is updated.
   
   


This is an automated message from the Apache Git Service.
To respond to the message, please log on GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


With regards,
Apache Git Services

[GitHub] anirudh2290 opened a new pull request #9869: Exception handling documentation

[GitBox]

anirudh2290 opened a new pull request #9869: Exception handling documentation
URL: https://github.com/apache/incubator-mxnet/pull/9869
 
 
   ## Description ##
   Adds documentation for exception handling. Requires #9795 to be merged


This is an automated message from the Apache Git Service.
To respond to the message, please log on GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


With regards,
Apache Git Services

[incubator-mxnet-site] 01/01: Merge pull request #52 from thinksanky/release_news_1.1.0

[liuyizhi]

This is an automated email from the ASF dual-hosted git repository.

liuyizhi pushed a commit to branch asf-site
in repository https://gitbox.apache.org/repos/asf/incubator-mxnet-site.git

commit a402a94223482212a6302d276f19aa15cfe9aacd
Merge: 876167c dde3caf
Author: Yizhi Liu 
AuthorDate: Thu Feb 22 22:57:50 2018 -0800

Merge pull request #52 from thinksanky/release_news_1.1.0

fixed the release number to 1.1.0

 index.html | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)

-- 
To stop receiving notification emails like this one, please contact
liuyi...@apache.org.

[GitHub] thinksanky opened a new pull request #52: fixed the release number to 1.1.0

[GitBox]

thinksanky opened a new pull request #52: fixed the release number to 1.1.0
URL: https://github.com/apache/incubator-mxnet-site/pull/52
 
 
   ## Description ##
   Updated the release number from 1.0 from 1.1.0


This is an automated message from the Apache Git Service.
To respond to the message, please log on GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


With regards,
Apache Git Services

[GitHub] dabraude opened a new issue #9868: MKL and CMake

[GitBox]

dabraude opened a new issue #9868: MKL and CMake
URL: https://github.com/apache/incubator-mxnet/issues/9868
 
 
   Looking at the CMake with MKL I noticed there is an option there called 
`USE_MKLML_MKL` which has the same docstring as `USE_MKLDNN` ; from what I can 
tell it is there for support with mshadow, would it not be safer to set 
`USE_MKLML_MKL` to be the same as `USE_MKLDNN` ?  Alternatively the docstring 
could indicate one is for mxnet and the other mshadow.
   
   


This is an automated message from the Apache Git Service.
To respond to the message, please log on GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


With regards,
Apache Git Services

[GitHub] juliusshufan opened a new pull request #9867: Using "uniform" Xavier strategy to initialize the weight for VGG network

[GitBox]

juliusshufan opened a new pull request #9867: Using "uniform" Xavier strategy 
to initialize the weight for VGG network
URL: https://github.com/apache/incubator-mxnet/pull/9867
 
 
   ## Description ##
   This PR provide a potential solution for issue 
https://github.com/apache/incubator-mxnet/issues/9866
   For detailed information, please check the issue.  
   
   ## Checklist ##
   ### Essentials ###
   - [ ] Passed code style checking (`make lint`)
   - [ ] Changes are complete (i.e. I finished coding on this PR)
   - [ ] All changes have test coverage:
   - Unit tests are added for small changes to verify correctness (e.g. adding 
a new operator)
   - Nightly tests are added for complicated/long-running ones (e.g. changing 
distributed kvstore)
   - Build tests will be added for build configuration changes (e.g. adding a 
new build option with NCCL)
   - [ ] Code is well-documented: 
   - For user-facing API changes, API doc string has been updated. 
   - For new C++ functions in header files, their functionalities and arguments 
are documented. 
   - For new examples, README.md is added to explain the what the example does, 
the source of the dataset, expected performance on test set and reference to 
the original paper if applicable
   - [ ] To the my best knowledge, examples are either not affected by this 
change, or have been fixed to be compatible with this change
   
   ### Changes ###
   example/image-classification/common/fit.py
   
   ## Comments ##
   This PR has been verified on Nvidia P40 and CPU machine
   


This is an automated message from the Apache Git Service.
To respond to the message, please log on GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


With regards,
Apache Git Services

[GitHub] juliusshufan opened a new issue #9866: The default weight initialization strategy makes the VGG network difficult to converge when utilizing examples under 'example/image-classification'

[GitBox]

juliusshufan opened a new issue #9866: The default weight initialization 
strategy makes the VGG network difficult to converge when utilizing examples 
under 'example/image-classification'
URL: https://github.com/apache/incubator-mxnet/issues/9866
 
 
   ## Description
   On trying to train CiFAR-10 dataset with VGG16, I noticed the initial 
learning rate has to be set a very small value to make the training converging. 
According the implementation of fit.py under 
example/image-classification/common, the Xavier with Gaussian random type is 
set as the default weight initializing strategy.
   
   On a NVIDIA P40 GPU, with **CUDA-9.1** and **openblas**, I executed the 
following comparison tests:
   Test 1: Initial LR = 0.01  Xavier with **_Gaussian_** random type
   Test 2: Initial LR = 0.00025Xavier with **_Gaussian_** random type
   Test 3: Initial LR = 0.01  Xavier with **_uniform_** random type
   
   The following two figures are corresponding to the comparison of the trends 
of training top-1 accuracy and cross-entropy loss. The curves in blue, green, 
purple are corresponding to test 1, 2, 3 respectively.
   
   
![image](https://user-images.githubusercontent.com/33112206/36579904-aae2350e-18a0-11e8-89aa-fc7e3876c0c3.png)
   
![image](https://user-images.githubusercontent.com/33112206/36579911-bdbcf1d2-18a0-11e8-8e02-d25a057bc56c.png)
   
   It can be observed, with same initial LR (0.01), the training with 
Xavier-Gaussian will not converge.
   Similar test results can also be observed when the dataset changed to 
CiFAR-100. (Retrieved from http://data.mxnet.io/data/cifar100.zip)  
   
   ## Environment info (Required)
   HW: NVIDIA P40
   SW:  MxNet master branch, with CUDA-v9.1, CUDNN v7.1 and openblas 2.1

   What to do:
   1. Download the diagnosis script from https://github.com/juliusshufan/mxnet 
   2. Run the three python script correspond to test 1, 2, 3
   (Note: The test script modified from the original train_cifar10.py coming 
with MXNET examples)
   
   ## Build info (Required if built from source)
   Compiler: gcc 4.8.5
   MXNet commit hash: cea8d2f4024a5e5d9d9edf43e42be130f10c7c27
   
   Build config:
   make -j($nproc) USE_OPENCV=1 USE_CUDA=1 USE_CUDNN=1 USE_BLAS=openblas 
USE_CUDA_PATH=/usr/local/cuda-9.1
   
   ## What have you tried to solve it?
   Using the Xavier with uniform random, but not the Gaussian random.
   


This is an automated message from the Apache Git Service.
To respond to the message, please log on GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


With regards,
Apache Git Services

[GitHub] zheng-da commented on issue #9862: Fix a race condition in converting data layouts in MKLDNN.

[GitBox]

zheng-da commented on issue #9862: Fix a race condition in converting data 
layouts in MKLDNN.
URL: https://github.com/apache/incubator-mxnet/pull/9862#issuecomment-367913772
 
 
   it seems the current modification still can't get rid of all race conditions 
in the code. the reason is that we want to reorder the data of weight arrays in 
place so that we can avoid reordering them again during inference. Right now, 
if I ran the tests hundreds of times, I might be able to see inconsistency 
once. @piiswrong and I have discussed a solution to get rid of remaining race 
conditions. it should work.


This is an automated message from the Apache Git Service.
To respond to the message, please log on GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


With regards,
Apache Git Services

[GitHub] pengzhao-intel commented on issue #9744: Performance regression when OMP_NUM_THREADS environment variable is not set

[GitBox]

pengzhao-intel commented on issue #9744: Performance regression when 
OMP_NUM_THREADS environment variable is not set
URL: 
https://github.com/apache/incubator-mxnet/issues/9744#issuecomment-367911898
 
 
   The OMP works now from the C code and I think KMP should be.
   
   There's C level API too, _int kmp_set_affinity_ in OpenMP, also on this page:
   https://software.intel.com/en-us/node/522691
   


This is an automated message from the Apache Git Service.
To respond to the message, please log on GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


With regards,
Apache Git Services

[GitHub] pengzhao-intel commented on issue #9744: Performance regression when OMP_NUM_THREADS environment variable is not set

[GitBox]

pengzhao-intel commented on issue #9744: Performance regression when 
OMP_NUM_THREADS environment variable is not set
URL: 
https://github.com/apache/incubator-mxnet/issues/9744#issuecomment-367911898
 
 
   There's C level API too, _int kmp_set_affinity_ in OpenMP. 
   The OMP works now and I think KMP should be.
   
   Also on this page:
   https://software.intel.com/en-us/node/522691
   


This is an automated message from the Apache Git Service.
To respond to the message, please log on GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


With regards,
Apache Git Services

[GitHub] pengzhao-intel commented on issue #9744: Performance regression when OMP_NUM_THREADS environment variable is not set

[GitBox]

pengzhao-intel commented on issue #9744: Performance regression when 
OMP_NUM_THREADS environment variable is not set
URL: 
https://github.com/apache/incubator-mxnet/issues/9744#issuecomment-367911898
 
 
   There're C level API too, _int kmp_set_affinity_ in openmp. 
   The OMP works now and I think KMP should be.
   
   Also in this page:
   https://software.intel.com/en-us/node/522691
   


This is an automated message from the Apache Git Service.
To respond to the message, please log on GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


With regards,
Apache Git Services

[GitHub] pengzhao-intel commented on issue #9744: Performance regression when OMP_NUM_THREADS environment variable is not set

[GitBox]

pengzhao-intel commented on issue #9744: Performance regression when 
OMP_NUM_THREADS environment variable is not set
URL: 
https://github.com/apache/incubator-mxnet/issues/9744#issuecomment-367911898
 
 
   There're C level API too, _int kmp_set_affinity_ in openmp.
   Also in this page:
   https://software.intel.com/en-us/node/522691
   


This is an automated message from the Apache Git Service.
To respond to the message, please log on GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


With regards,
Apache Git Services

[GitHub] ashokei commented on issue #9810: remove MKL_EXPERIMENTAL and update make files for MKL-DNN

[GitBox]

ashokei commented on issue #9810: remove MKL_EXPERIMENTAL and update make files 
for MKL-DNN
URL: https://github.com/apache/incubator-mxnet/pull/9810#issuecomment-367908298
 
 
   @marcoabreu can you please review requested Jenkins file changes; This PR is 
primarily targeting at removing MKL_EXPERIMENTAL and minor changes to regular 
makefiles; we may do a separate PR for cmake mkldnn build as required. thanks!


This is an automated message from the Apache Git Service.
To respond to the message, please log on GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


With regards,
Apache Git Services

[GitHub] zheng-da commented on a change in pull request #9862: Fix a race condition in converting data layouts in MKLDNN.

[GitBox]

zheng-da commented on a change in pull request #9862: Fix a race condition in 
converting data layouts in MKLDNN.
URL: https://github.com/apache/incubator-mxnet/pull/9862#discussion_r170165614
 
 

 ##
 File path: src/ndarray/ndarray.cc
 ##
 @@ -1017,6 +1017,7 @@ inline void CopyFromToDnsImpl(const NDArray& from, const 
NDArray& to, RunContext
 // with Copy().
 NDArray tmp_from = from;
 if (tmp_from.IsMKLDNNData()) {
+  // TODO(zhengda) tmp_from should be cached.
 
 Review comment:
   const can have a lot of definitions. I don't think it's a bad design. here 
const means the values of the array are always the same, even though its layout 
may be changed. if we strictly follow the rule that const function doesn't 
modify data in an array, the code would become unmanagable. In fact, the data 
in a const array in MXNet is modified all the time. 


This is an automated message from the Apache Git Service.
To respond to the message, please log on GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


With regards,
Apache Git Services

[GitHub] cjolivier01 commented on issue #9744: Performance regression when OMP_NUM_THREADS environment variable is not set

[GitBox]

cjolivier01 commented on issue #9744: Performance regression when 
OMP_NUM_THREADS environment variable is not set
URL: 
https://github.com/apache/incubator-mxnet/issues/9744#issuecomment-367907988
 
 
   i don?t set the environment variable. I doubt it would
   work even if i did, from within the code. So there?s nothing I can do,
   then, right? User would need to set?
   
   I recall dumping the default behavior of KMP and i remember it showing that
   the OMP threads were indeed staggered as in your picture.  I?ll check again.
   
   On Thu, Feb 22, 2018 at 8:13 PM PatricZhao  wrote:
   
   > Normally, we set KMP_AFFILITY with OMP_NUM_THREADS together.
   > And when the OMP threads are invoked, it will be used.
   >
   > ?
   > You are receiving this because you were mentioned.
   > Reply to this email directly, view it on GitHub
   > 
,
   > or mute the thread
   > 

   > .
   >
   


This is an automated message from the Apache Git Service.
To respond to the message, please log on GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


With regards,
Apache Git Services

[GitHub] pengzhao-intel commented on issue #9744: Performance regression when OMP_NUM_THREADS environment variable is not set

[GitBox]

pengzhao-intel commented on issue #9744: Performance regression when 
OMP_NUM_THREADS environment variable is not set
URL: 
https://github.com/apache/incubator-mxnet/issues/9744#issuecomment-367904801
 
 
   Normally, we set KMP_AFFILITY with OMP_NUM_THREADS together. 
   And when the OMP threads are invoked, it will be used. 


This is an automated message from the Apache Git Service.
To respond to the message, please log on GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


With regards,
Apache Git Services

[GitHub] cjolivier01 commented on a change in pull request #9862: Fix a race condition in converting data layouts in MKLDNN.

[GitBox]

cjolivier01 commented on a change in pull request #9862: Fix a race condition 
in converting data layouts in MKLDNN.
URL: https://github.com/apache/incubator-mxnet/pull/9862#discussion_r170162245
 
 

 ##
 File path: src/operator/nn/mkldnn/mkldnn_base.cc
 ##
 @@ -270,9 +270,24 @@ void FallBackCompute(FCompute fn, const nnvm::NodeAttrs 
,
  const std::vector ,
  const std::vector ) {
   std::vector in_blobs(inputs.size());
+  std::vector in_bufs;
   for (size_t i = 0; i < in_blobs.size(); i++) {
+// If the input data isn't stored in the default format, we shouldn't
+// call data() directly, which will change the layout of the NDArray.
+// Instead, we should save the converted data in another NDArray.
+// TODO(zhengda) we should use temp space to save the converted data.
+if (inputs[i].IsDefaultData()) {
   in_blobs[i] = inputs[i].data();
+} else {
+  in_bufs.emplace_back(inputs[i].shape(), inputs[i].ctx(),
 
 Review comment:
   implementations vary, but this gives a general idea:
   
   http://www.drdobbs.com/c-made-easier-how-vectors-grow/184401375


This is an automated message from the Apache Git Service.
To respond to the message, please log on GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


With regards,
Apache Git Services

[GitHub] cjolivier01 commented on a change in pull request #9862: Fix a race condition in converting data layouts in MKLDNN.

[GitBox]

cjolivier01 commented on a change in pull request #9862: Fix a race condition 
in converting data layouts in MKLDNN.
URL: https://github.com/apache/incubator-mxnet/pull/9862#discussion_r170161984
 
 

 ##
 File path: src/ndarray/ndarray.cc
 ##
 @@ -1017,6 +1017,7 @@ inline void CopyFromToDnsImpl(const NDArray& from, const 
NDArray& to, RunContext
 // with Copy().
 NDArray tmp_from = from;
 if (tmp_from.IsMKLDNNData()) {
+  // TODO(zhengda) tmp_from should be cached.
 
 Review comment:
   if it doesn?t require const_cast<>, then it?s perfectly legal.


This is an automated message from the Apache Git Service.
To respond to the message, please log on GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


With regards,
Apache Git Services

[GitHub] cjolivier01 commented on issue #9862: Fix a race condition in converting data layouts in MKLDNN.

[GitBox]

cjolivier01 commented on issue #9862: Fix a race condition in converting data 
layouts in MKLDNN.
URL: https://github.com/apache/incubator-mxnet/pull/9862#issuecomment-367902985
 
 
   if you run the test in a loop (from python) 1000 times, does it fail?


This is an automated message from the Apache Git Service.
To respond to the message, please log on GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


With regards,
Apache Git Services

[GitHub] cjolivier01 commented on a change in pull request #9862: Fix a race condition in converting data layouts in MKLDNN.

[GitBox]

cjolivier01 commented on a change in pull request #9862: Fix a race condition 
in converting data layouts in MKLDNN.
URL: https://github.com/apache/incubator-mxnet/pull/9862#discussion_r170161424
 
 

 ##
 File path: src/operator/nn/mkldnn/mkldnn_base.cc
 ##
 @@ -270,9 +270,24 @@ void FallBackCompute(FCompute fn, const nnvm::NodeAttrs 
,
  const std::vector ,
  const std::vector ) {
   std::vector in_blobs(inputs.size());
+  std::vector in_bufs;
   for (size_t i = 0; i < in_blobs.size(); i++) {
+// If the input data isn't stored in the default format, we shouldn't
+// call data() directly, which will change the layout of the NDArray.
+// Instead, we should save the converted data in another NDArray.
+// TODO(zhengda) we should use temp space to save the converted data.
+if (inputs[i].IsDefaultData()) {
   in_blobs[i] = inputs[i].data();
+} else {
+  in_bufs.emplace_back(inputs[i].shape(), inputs[i].ctx(),
 
 Review comment:
   when you add items to a vector, when it runs out of space for the items 
(which generally happens on the second item sometimes), it allocates memory for 
2x the current number of items and then copies all of the items (copy 
constructor or move, depending on a number of things), into to the new space 
and then frees the old space. this can happen over and over again, and the 
whole reallocate and copy can be very expensive. they tend to even show up as 
hotspots in vtune. so if you know how many items a vector may have, it?s always 
a good idea to call reserve() because otherwise you are guaranteed a lot of 
extra reallocing And copying.


This is an automated message from the Apache Git Service.
To respond to the message, please log on GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


With regards,
Apache Git Services

[GitHub] feevos commented on issue #9822: gluon HybridBlock wrapper of constant nd.array, is it possible?

[GitBox]

feevos commented on issue #9822: gluon HybridBlock wrapper of constant 
nd.array, is it possible?
URL: 
https://github.com/apache/incubator-mxnet/issues/9822#issuecomment-367901846
 
 
   Thanks @jmacglashan , based on your suggestions I tried the following 
solution: 
   
   
   I am trying to create a ```HybridBlock``` custom Convolutional layer.  I am 
creating a custom initializer and then I create inside HybridBlock a custom 
parameter  ```self.bijkl``` that is initialized with the custom initializer 
```CustomInit```. The full example is here: 
   
   Essentials:
   ```Python
   import mxnet as mx
   from mxnet import nd, gluon
   from mxnet.gluon import HybridBlock
   ```
   
   
   ```Python
   from scipy.interpolate import BSpline
   
   
   @mx.init.register
   class CustomInit(mx.initializer.Initializer):
   
   def __init__(self,
kernel_size = 3, 
kernel_effective_size = 5, 
**kwards):
   mx.initializer.Initializer.__init__(self,**kwards)
   
# A trivial constant tensor
   self.Bijkl = 
nd.random_uniform(shape=[kernel_size,kernel_size,kernel_effective_size,kernel_effective_size])
   
   

   def _init_weight(self,name,arr):
   arr[:] = self.Bijkl
   
   
   class Conv2DS(HybridBlock):
   def __init__(self,  nfilters, nchannels=0, kernel_size = 3, 
kernel_effective_size = 5, use_bias = True, padding = (0,0), **kwards):
   HybridBlock.__init__(self,**kwards)
   
   self.nchannels = nchannels
   self.nfilters = nfilters
   self.kernel_size = kernel_size
   self.kernel_eff = kernel_effective_size
   self.use_bias = use_bias
   self.pad = padding
  
   
   with self.name_scope():
   
   
   self.weight = self.params.get(
   'weight',allow_deferred_init=True,
   shape=(nfilters,nchannels,kernel_size,kernel_size))
   
# This is the custom tensor I need to create, a constant. 
   self.Bijkl = self.params.get(
   'bijkl',allow_deferred_init=True,
   init = CustomInit(self.kernel_size, self.kernel_eff),
   grad_req='null',
   
shape=(kernel_size,kernel_size,kernel_effective_size,kernel_effective_size))
   
   
   if self.use_bias:
   self.bias = self.params.get(
   'bias',allow_deferred_init=True,
   init = mx.init.Zero(),
   shape=(self.nfilters,))
   
   
   def hybrid_forward(self,F,_x):
   # These finalize deferring the input shape 
   # 

   #self.weight.shape = 
(self.nfilters,_x.shape[1],self.kernel_size,self.kernel_size)
   #self.weight._finish_deferred_init()
   # 

 
   
   
   weight = F.sum(F.dot(self.weight.data() , 
self.Bijkl.data()),axis=[2,3])
   
   if self.use_bias:
   conv = F.Convolution(data=_x,
 weight=weight,
 bias=self.bias.data(),
 num_filter=self.nfilters,
 kernel=[self.kernel_eff,self.kernel_eff],
 pad = self.pad)
   
   else : 
   conv = F.Convolution(data=_x,
 weight=weight,
 no_bias=True,
 num_filter=self.nfilters,
 kernel=[self.kernel_eff,self.kernel_eff],
 pad = self.pad)
   
   
   return conv
   
   ```
   
   I can initialize my layer: 
   
   ```Python 
   nchannels = 8
   nfilters = 16
   dim = 128
   
   mynet = Conv2DS(nfilters, kernel_size=5, kernel_effective_size= 15, 
use_bias=False)
   
   mynet.initialize(mx.init.Xavier(),ctx=mx.gpu())
   ```
   
   but when I run a single pass I get the following error: 
   ```Python
   xx = nd.random_uniform(shape = [4,nchannels,dim,dim],ctx=mx.gpu())
   temp = mynet (xx)
   ```
   ## Error code:
   
   ```Python
   ---
   TypeError Traceback (most recent call last)
in ()
   > 1 temp = mynet (xx)
   
   /home/dia021/anaconda2/lib/python2.7/site-packages/mxnet/gluon/block.pyc in 
__call__(self, *args)
   358 def __call__(self, *args):
   359 """Calls forward. Only accepts positional arguments."""
   --> 360 return self.forward(*args)
   361 
   

[GitHub] cjolivier01 commented on issue #9744: Performance regression when OMP_NUM_THREADS environment variable is not set

[GitBox]

cjolivier01 commented on issue #9744: Performance regression when 
OMP_NUM_THREADS environment variable is not set
URL: 
https://github.com/apache/incubator-mxnet/issues/9744#issuecomment-367901652
 
 
   how to set this programmatically? i?d imagine that by the time this code
   executes, the omp library may have already checked the environment variable
   (as it does with OMP_NUM_THREADS)? or can it be set at any time?
   
   On Thu, Feb 22, 2018 at 4:49 PM PatricZhao  wrote:
   
   > Thanks, @cjolivier01 , I mean we set this
   > environment in the code as OMP_NUM_THREADS.
   >
   > Actually, this setting is useful when hyperthreading is ON and we only
   > want to run each thread in a physical core as below picture shown
   > (KMP_AFFINITY=granularity=fine,compact,1,0).
   >
   > So, this environment will guide how to map thread into the physical core.
   > https://software.intel.com/en-us/node/522691
   >
   > [image: guid-80fda344-d8c1-4fea-b268-94f685d19c4c-imageid
   > 0abe3e2e-7a5a-4072-b22f-da7f72eb9a18]
   > 

   > *KMP_AFFINITY=granularity=fine,compact,1,0*
   >
   > ?
   > You are receiving this because you were mentioned.
   > Reply to this email directly, view it on GitHub
   > 
,
   > or mute the thread
   > 

   > .
   >
   


This is an automated message from the Apache Git Service.
To respond to the message, please log on GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


With regards,
Apache Git Services

[GitHub] anirudh2290 commented on issue #9813: Unable to save gluon model to symbolic network : neural style

[GitBox]

anirudh2290 commented on issue #9813: Unable to save gluon model to symbolic 
network : neural style
URL: 
https://github.com/apache/incubator-mxnet/issues/9813#issuecomment-367901245
 
 
   do you have a validation dataset to test the model ?


This is an automated message from the Apache Git Service.
To respond to the message, please log on GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


With regards,
Apache Git Services

[GitHub] xiaoheizi123 commented on issue #7162: Cant convert caffe model to mxnet

[GitBox]

xiaoheizi123 commented on issue #7162: Cant convert caffe model to mxnet
URL: 
https://github.com/apache/incubator-mxnet/issues/7162#issuecomment-367900241
 
 
   I also cannot convert resnet-50 using **python convert_caffe_modelzoo.py 
resnet-50**
   
   `converting scale layer, beta shape = (2048,), gamma shape = (2048,)
   skipping layer res5c of type Eltwise
   skipping layer res5c_relu of type ReLU
   skipping layer pool5 of type Pooling
   converting layer fc1000, wmat shape = (1000, 2048), bias shape = (1000,)
   skipping layer prob of type Softmax
   
/home/users/rui.zheng/anaconda2/lib/python2.7/site-packages/mxnet/module/base_module.py:65:
 UserWarning: Data provided by label_shapes don't match names specified by 
label_names ([] vs. ['prob_label'])
 warnings.warn(msg)
   Traceback (most recent call last):
 File "convert_caffe_modelzoo.py", line 144, in 
   fname, _ = convert_caffe_model(args.model_name, 
model_meta_info[args.model_name])
 File "convert_caffe_modelzoo.py", line 133, in convert_caffe_model
   convert_mean(mean, [mx_mean])
 File 
"/home/users/rui.zheng/works/incubator-mxnet/tools/caffe_converter/convert_mean.py",
 line 44, in convert_mean
   mean_blob.ParseFromString(f.read())
   google.protobuf.message.DecodeError: Error parsing message
   `
   f.read() is ok, but mean_blob is None when print it?
   I also read 
[https://github.com/apache/incubator-mxnet/blob/430ea7bfbbda67d993996d81c7fd44d3a20ef846/docs/how_to/caffe.md](url),
 and do what **Converting Caffe trained models to MXNet**  part shows, but it 
is still wrong, what should I do? 


This is an automated message from the Apache Git Service.
To respond to the message, please log on GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


With regards,
Apache Git Services

[GitHub] sxjscience commented on issue #9843: Flaky test_random.test_exponential_generator @ Python2: MKLDNN-CPU

[GitBox]

sxjscience commented on issue #9843: Flaky 
test_random.test_exponential_generator @ Python2: MKLDNN-CPU
URL: 
https://github.com/apache/incubator-mxnet/issues/9843#issuecomment-367879896
 
 
   @marcoabreu A wrongly implemented random number generator would typically 
have success_rate = 0, i.e, producing p < 0.05 for all the 5 runs. However, 
since this tests the randomness, it's really tricky to set the threshold and my 
previous design is to force the success rate to be at least 25%.


This is an automated message from the Apache Git Service.
To respond to the message, please log on GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


With regards,
Apache Git Services

[GitHub] larroy commented on a change in pull request #9862: Fix a race condition in converting data layouts in MKLDNN.

[GitBox]

larroy commented on a change in pull request #9862: Fix a race condition in 
converting data layouts in MKLDNN.
URL: https://github.com/apache/incubator-mxnet/pull/9862#discussion_r170141588
 
 

 ##
 File path: src/ndarray/ndarray.cc
 ##
 @@ -1017,6 +1017,7 @@ inline void CopyFromToDnsImpl(const NDArray& from, const 
NDArray& to, RunContext
 // with Copy().
 NDArray tmp_from = from;
 if (tmp_from.IsMKLDNNData()) {
+  // TODO(zhengda) tmp_from should be cached.
 
 Review comment:
   I'm not familiar with this code. But why are all all these operations 
mutating the array in a function that is supposed to be const?  I think that is 
really bad design that a call to a const function (data) is reodering memory in 
the tensor. Why hasn't this come up in the code review?


This is an automated message from the Apache Git Service.
To respond to the message, please log on GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


With regards,
Apache Git Services

[GitHub] szha closed pull request #9863: Fix broken website build and update version number

[GitBox]

szha closed pull request #9863: Fix broken website build and update version 
number
URL: https://github.com/apache/incubator-mxnet/pull/9863
 
 
   

This is a PR merged from a forked repository.
As GitHub hides the original diff on merge, it is displayed below for
the sake of provenance:

As this is a foreign pull request (from a fork), the diff is supplied
below (as it won't show otherwise due to GitHub magic):

diff --git a/docs/build_version_doc/build_all_version.sh 
b/docs/build_version_doc/build_all_version.sh
index bd3758b4b2..6a37815fd5 100755
--- a/docs/build_version_doc/build_all_version.sh
+++ b/docs/build_version_doc/build_all_version.sh
@@ -21,7 +21,10 @@
 # Built files are stored in $built
 # Version numbers are stored in $tag_list.
 # Version numbers are ordered from latest to old and final one is master.
-tag_list="1.0.0 0.12.1 0.12.0 0.11.0 master"
+set -e
+set -x
+
+tag_list="1.1.0 1.0.0 0.12.1 0.12.0 0.11.0 master"
 
 mxnet_url="https://github.com/apache/incubator-mxnet.git;
 mxnet_folder="apache_mxnet"
diff --git a/docs/build_version_doc/build_doc.sh 
b/docs/build_version_doc/build_doc.sh
index b13fa5c191..eefc81e362 100755
--- a/docs/build_version_doc/build_doc.sh
+++ b/docs/build_version_doc/build_doc.sh
@@ -16,7 +16,8 @@
 # KIND, either express or implied.  See the License for the
 # specific language governing permissions and limitations
 # under the License.
-
+set -e
+set -x
 
 web_url="$1"
 web_folder="VersionedWeb"
@@ -57,7 +58,7 @@ then
 cat $tag_list_file
 tests/ci_build/ci_build.sh doc python docs/build_version_doc/AddVersion.py 
--file_path "docs/_build/html/" --current_version "$latest_tag"
 tests/ci_build/ci_build.sh doc python 
docs/build_version_doc/AddPackageLink.py \
-  --file_path 
"docs/_build/html/get_started/install.html" --current_version "$latest_tag"
+  --file_path 
"docs/_build/html/install/index.html" --current_version "$latest_tag"
 cp -a "docs/_build/html/." "$local_build"
 cp $tag_list_file "$local_build/tag.txt"
 rm -rf "$web_folder/.git"


 


This is an automated message from the Apache Git Service.
To respond to the message, please log on GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


With regards,
Apache Git Services

[incubator-mxnet] branch master updated: Fix broken website build, add latest version number and made build scripts fail fast (#9863)

[zhasheng]

This is an automated email from the ASF dual-hosted git repository.

zhasheng pushed a commit to branch master
in repository https://gitbox.apache.org/repos/asf/incubator-mxnet.git


The following commit(s) were added to refs/heads/master by this push:
 new cea8d2f  Fix broken website build, add latest version number and made 
build scripts fail fast (#9863)
cea8d2f is described below

commit cea8d2f4024a5e5d9d9edf43e42be130f10c7c27
Author: Marco de Abreu 
AuthorDate: Fri Feb 23 02:14:16 2018 +0100

Fix broken website build, add latest version number and made build scripts 
fail fast (#9863)
---
 docs/build_version_doc/build_all_version.sh | 5 -
 docs/build_version_doc/build_doc.sh | 5 +++--
 2 files changed, 7 insertions(+), 3 deletions(-)

diff --git a/docs/build_version_doc/build_all_version.sh 
b/docs/build_version_doc/build_all_version.sh
index bd3758b..6a37815 100755
--- a/docs/build_version_doc/build_all_version.sh
+++ b/docs/build_version_doc/build_all_version.sh
@@ -21,7 +21,10 @@
 # Built files are stored in $built
 # Version numbers are stored in $tag_list.
 # Version numbers are ordered from latest to old and final one is master.
-tag_list="1.0.0 0.12.1 0.12.0 0.11.0 master"
+set -e
+set -x
+
+tag_list="1.1.0 1.0.0 0.12.1 0.12.0 0.11.0 master"
 
 mxnet_url="https://github.com/apache/incubator-mxnet.git;
 mxnet_folder="apache_mxnet"
diff --git a/docs/build_version_doc/build_doc.sh 
b/docs/build_version_doc/build_doc.sh
index b13fa5c..eefc81e 100755
--- a/docs/build_version_doc/build_doc.sh
+++ b/docs/build_version_doc/build_doc.sh
@@ -16,7 +16,8 @@
 # KIND, either express or implied.  See the License for the
 # specific language governing permissions and limitations
 # under the License.
-
+set -e
+set -x
 
 web_url="$1"
 web_folder="VersionedWeb"
@@ -57,7 +58,7 @@ then
 cat $tag_list_file
 tests/ci_build/ci_build.sh doc python docs/build_version_doc/AddVersion.py 
--file_path "docs/_build/html/" --current_version "$latest_tag"
 tests/ci_build/ci_build.sh doc python 
docs/build_version_doc/AddPackageLink.py \
-  --file_path 
"docs/_build/html/get_started/install.html" --current_version "$latest_tag"
+  --file_path 
"docs/_build/html/install/index.html" --current_version "$latest_tag"
 cp -a "docs/_build/html/." "$local_build"
 cp $tag_list_file "$local_build/tag.txt"
 rm -rf "$web_folder/.git"

-- 
To stop receiving notification emails like this one, please contact
zhash...@apache.org.

[GitHub] larroy commented on a change in pull request #9862: Fix a race condition in converting data layouts in MKLDNN.

[GitBox]

larroy commented on a change in pull request #9862: Fix a race condition in 
converting data layouts in MKLDNN.
URL: https://github.com/apache/incubator-mxnet/pull/9862#discussion_r170141588
 
 

 ##
 File path: src/ndarray/ndarray.cc
 ##
 @@ -1017,6 +1017,7 @@ inline void CopyFromToDnsImpl(const NDArray& from, const 
NDArray& to, RunContext
 // with Copy().
 NDArray tmp_from = from;
 if (tmp_from.IsMKLDNNData()) {
+  // TODO(zhengda) tmp_from should be cached.
 
 Review comment:
   I'm not familiar with this code. But why are all all these operations 
mutating the array in a function that is supposed to be const?


This is an automated message from the Apache Git Service.
To respond to the message, please log on GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


With regards,
Apache Git Services

[GitHub] zheng-da commented on issue #9862: Fix a race condition in converting data layouts in MKLDNN.

[GitBox]

zheng-da commented on issue #9862: Fix a race condition in converting data 
layouts in MKLDNN.
URL: https://github.com/apache/incubator-mxnet/pull/9862#issuecomment-367876047
 
 
   It's very difficult to reproduce a race condition in a deterministic way if 
it's possible. 


This is an automated message from the Apache Git Service.
To respond to the message, please log on GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


With regards,
Apache Git Services

[GitHub] larroy commented on a change in pull request #9862: Fix a race condition in converting data layouts in MKLDNN.

[GitBox]

larroy commented on a change in pull request #9862: Fix a race condition in 
converting data layouts in MKLDNN.
URL: https://github.com/apache/incubator-mxnet/pull/9862#discussion_r170139781
 
 

 ##
 File path: src/operator/nn/mkldnn/mkldnn_base.cc
 ##
 @@ -270,9 +270,24 @@ void FallBackCompute(FCompute fn, const nnvm::NodeAttrs 
,
  const std::vector ,
  const std::vector ) {
   std::vector in_blobs(inputs.size());
+  std::vector in_bufs;
   for (size_t i = 0; i < in_blobs.size(); i++) {
+// If the input data isn't stored in the default format, we shouldn't
+// call data() directly, which will change the layout of the NDArray.
+// Instead, we should save the converted data in another NDArray.
+// TODO(zhengda) we should use temp space to save the converted data.
+if (inputs[i].IsDefaultData()) {
   in_blobs[i] = inputs[i].data();
+} else {
+  in_bufs.emplace_back(inputs[i].shape(), inputs[i].ctx(),
 
 Review comment:
   It prevents reallocations, heap fragmentation, and contention if you know 
the size, it's definitely good to use reserve in advance.


This is an automated message from the Apache Git Service.
To respond to the message, please log on GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


With regards,
Apache Git Services

[GitHub] marcoabreu commented on issue #9862: Fix a race condition in converting data layouts in MKLDNN.

[GitBox]

marcoabreu commented on issue #9862: Fix a race condition in converting data 
layouts in MKLDNN.
URL: https://github.com/apache/incubator-mxnet/pull/9862#issuecomment-367872237
 
 
   Would it be possible to create a test that introduces a delay or uses a 
for-loop in order to force this race condition? 


This is an automated message from the Apache Git Service.
To respond to the message, please log on GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


With regards,
Apache Git Services

[GitHub] nswamy closed pull request #9851: Softsign op

[GitBox]

nswamy closed pull request #9851: Softsign op
URL: https://github.com/apache/incubator-mxnet/pull/9851
 
 
   

This is a PR merged from a forked repository.
As GitHub hides the original diff on merge, it is displayed below for
the sake of provenance:

As this is a foreign pull request (from a fork), the diff is supplied
below (as it won't show otherwise due to GitHub magic):

diff --git a/src/operator/mshadow_op.h b/src/operator/mshadow_op.h
index af7ef513f1..1d4284e1ac 100644
--- a/src/operator/mshadow_op.h
+++ b/src/operator/mshadow_op.h
@@ -111,6 +111,10 @@ MXNET_UNARY_MATH_OP(sigmoid, 1.0f / (1.0f + 
math::exp(-a)));
 
 MXNET_UNARY_MATH_OP(sigmoid_grad, math::id(a) * (1.0f - math::id(a)));
 
+MXNET_UNARY_MATH_OP(softsign, a / (1.0f + math::fabs(a)));
+
+MXNET_UNARY_MATH_OP(softsign_grad, 1.0f /  math::sqr(1.0f + math::fabs(a)));
+
 MXNET_UNARY_MATH_OP_NC(relu, a > DType(0) ? a : DType(0));
 
 MXNET_UNARY_MATH_OP_NC(relu_grad, a > DType(0) ? DType(1) : DType(0));
diff --git a/src/operator/nn/activation-inl.h b/src/operator/nn/activation-inl.h
index a440f97e13..89a369c671 100644
--- a/src/operator/nn/activation-inl.h
+++ b/src/operator/nn/activation-inl.h
@@ -47,7 +47,7 @@ namespace activation {
 enum ActivationOpInputs {kData};
 enum ActivationOpOutputs {kOut};
 enum ActivationOpResource {kTempSpace};
-enum ActivationOpType {kReLU, kSigmoid, kTanh, kSoftReLU};
+enum ActivationOpType {kReLU, kSigmoid, kTanh, kSoftReLU, kSoftSign};
 }  // activation
 
 struct ActivationParam : public dmlc::Parameter {
@@ -59,6 +59,7 @@ struct ActivationParam : public 
dmlc::Parameter {
 .add_enum("sigmoid", activation::kSigmoid)
 .add_enum("tanh", activation::kTanh)
 .add_enum("softrelu", activation::kSoftReLU)
+.add_enum("softsign", activation::kSoftSign)
 .describe("Activation function to be applied.");
   }
 
@@ -140,6 +141,10 @@ void ActivationComputeImpl(const ActivationParam , 
const OpContext ,
 ActivationForward(
 ctx, input, req, output);
 break;
+  case activation::kSoftSign:
+ActivationForward(
+ctx, input, req, output);
+break;
   default:
 LOG(FATAL) << "unknown activation type";
 }
@@ -168,6 +173,10 @@ void ActivationGradComputeImpl(const ActivationParam 
, const OpContext 
 ActivationBackward(
 ctx, out_grad, out_data, req, output);
 break;
+  case activation::kSoftSign:
+ActivationBackward(
+ctx, out_grad, out_data, req, output);
+break;
   default:
 LOG(FATAL) << "unknown activation type";
 }
diff --git a/src/operator/nn/activation.cc b/src/operator/nn/activation.cc
index 7b79a34a7b..89059321b6 100644
--- a/src/operator/nn/activation.cc
+++ b/src/operator/nn/activation.cc
@@ -149,6 +149,7 @@ The following activation functions are supported:
 - `sigmoid`: :math:`y = \frac{1}{1 + exp(-x)}`
 - `tanh`: Hyperbolic tangent, :math:`y = \frac{exp(x) - exp(-x)}{exp(x) + 
exp(-x)}`
 - `softrelu`: Soft ReLU, or SoftPlus, :math:`y = log(1 + exp(x))`
+- `softsign`: :math:`y = \frac{x}{1 + abs(x)}`
 
 )code" ADD_FILELINE)
 .set_attr_parser(ParamParser)
diff --git a/src/operator/operator_tune.cc b/src/operator/operator_tune.cc
index e0f8306565..c13f1ac2fa 100644
--- a/src/operator/operator_tune.cc
+++ b/src/operator/operator_tune.cc
@@ -213,6 +213,8 @@ 
IMPLEMENT_UNARY_WORKLOAD_FWD(mxnet::op::mshadow_op::reciprocal);  // NOLINT()
 IMPLEMENT_UNARY_WORKLOAD_BWD(mxnet::op::mshadow_op::reciprocal_grad);  // 
NOLINT()
 IMPLEMENT_UNARY_WORKLOAD_FWD(mxnet::op::mshadow_op::sigmoid);  // NOLINT()
 IMPLEMENT_UNARY_WORKLOAD_BWD(mxnet::op::mshadow_op::sigmoid_grad);  // NOLINT()
+IMPLEMENT_UNARY_WORKLOAD_FWD(mxnet::op::mshadow_op::softsign);  // NOLINT()
+IMPLEMENT_UNARY_WORKLOAD_BWD(mxnet::op::mshadow_op::softsign_grad);  // 
NOLINT()
 IMPLEMENT_UNARY_WORKLOAD_FWD(mxnet::op::mshadow_op::relu);  // NOLINT()
 IMPLEMENT_UNARY_WORKLOAD_BWD(mxnet::op::mshadow_op::relu_grad);  // NOLINT()
 IMPLEMENT_UNARY_WORKLOAD_FWD(mxnet::op::mshadow_op::tanh);  // NOLINT()
diff --git a/src/operator/tensor/elemwise_unary_op_basic.cc 
b/src/operator/tensor/elemwise_unary_op_basic.cc
index 95fd3bc7e3..acd8f7b23f 100644
--- a/src/operator/tensor/elemwise_unary_op_basic.cc
+++ b/src/operator/tensor/elemwise_unary_op_basic.cc
@@ -106,6 +106,23 @@ The storage type of ``sigmoid`` output is always dense
 
 MXNET_OPERATOR_REGISTER_BINARY_WITH_SPARSE_CPU(_backward_sigmoid,

unary_bwd);
+// softsign
+MXNET_OPERATOR_REGISTER_UNARY(softsign)
+MXNET_ADD_SPARSE_OP_ALIAS(softsign)
+.describe(R"code(Computes softsign of x element-wise.
+
+.. math::
+   y = x / (1 + abs(x))
+
+The storage 

[incubator-mxnet] branch master updated: Softsign Activation Function (#9851)

[nswamy]

This is an automated email from the ASF dual-hosted git repository.

nswamy pushed a commit to branch master
in repository https://gitbox.apache.org/repos/asf/incubator-mxnet.git


The following commit(s) were added to refs/heads/master by this push:
 new 6ebec87  Softsign Activation Function (#9851)
6ebec87 is described below

commit 6ebec87127a7dfb10e902df20282434a1612e35f
Author: Naveen Swamy 
AuthorDate: Thu Feb 22 16:54:21 2018 -0800

Softsign Activation Function (#9851)

* Add SoftSign Activation function
---
 src/operator/mshadow_op.h  |  4 
 src/operator/nn/activation-inl.h   | 11 ++-
 src/operator/nn/activation.cc  |  1 +
 src/operator/operator_tune.cc  |  2 ++
 src/operator/tensor/elemwise_unary_op_basic.cc | 17 +
 src/operator/tensor/elemwise_unary_op_basic.cu |  8 
 tests/python/unittest/test_operator.py | 19 +++
 7 files changed, 61 insertions(+), 1 deletion(-)

diff --git a/src/operator/mshadow_op.h b/src/operator/mshadow_op.h
index af7ef51..1d4284e 100644
--- a/src/operator/mshadow_op.h
+++ b/src/operator/mshadow_op.h
@@ -111,6 +111,10 @@ MXNET_UNARY_MATH_OP(sigmoid, 1.0f / (1.0f + 
math::exp(-a)));
 
 MXNET_UNARY_MATH_OP(sigmoid_grad, math::id(a) * (1.0f - math::id(a)));
 
+MXNET_UNARY_MATH_OP(softsign, a / (1.0f + math::fabs(a)));
+
+MXNET_UNARY_MATH_OP(softsign_grad, 1.0f /  math::sqr(1.0f + math::fabs(a)));
+
 MXNET_UNARY_MATH_OP_NC(relu, a > DType(0) ? a : DType(0));
 
 MXNET_UNARY_MATH_OP_NC(relu_grad, a > DType(0) ? DType(1) : DType(0));
diff --git a/src/operator/nn/activation-inl.h b/src/operator/nn/activation-inl.h
index a440f97..89a369c 100644
--- a/src/operator/nn/activation-inl.h
+++ b/src/operator/nn/activation-inl.h
@@ -47,7 +47,7 @@ namespace activation {
 enum ActivationOpInputs {kData};
 enum ActivationOpOutputs {kOut};
 enum ActivationOpResource {kTempSpace};
-enum ActivationOpType {kReLU, kSigmoid, kTanh, kSoftReLU};
+enum ActivationOpType {kReLU, kSigmoid, kTanh, kSoftReLU, kSoftSign};
 }  // activation
 
 struct ActivationParam : public dmlc::Parameter {
@@ -59,6 +59,7 @@ struct ActivationParam : public 
dmlc::Parameter {
 .add_enum("sigmoid", activation::kSigmoid)
 .add_enum("tanh", activation::kTanh)
 .add_enum("softrelu", activation::kSoftReLU)
+.add_enum("softsign", activation::kSoftSign)
 .describe("Activation function to be applied.");
   }
 
@@ -140,6 +141,10 @@ void ActivationComputeImpl(const ActivationParam , 
const OpContext ,
 ActivationForward(
 ctx, input, req, output);
 break;
+  case activation::kSoftSign:
+ActivationForward(
+ctx, input, req, output);
+break;
   default:
 LOG(FATAL) << "unknown activation type";
 }
@@ -168,6 +173,10 @@ void ActivationGradComputeImpl(const ActivationParam 
, const OpContext 
 ActivationBackward(
 ctx, out_grad, out_data, req, output);
 break;
+  case activation::kSoftSign:
+ActivationBackward(
+ctx, out_grad, out_data, req, output);
+break;
   default:
 LOG(FATAL) << "unknown activation type";
 }
diff --git a/src/operator/nn/activation.cc b/src/operator/nn/activation.cc
index 7b79a34..8905932 100644
--- a/src/operator/nn/activation.cc
+++ b/src/operator/nn/activation.cc
@@ -149,6 +149,7 @@ The following activation functions are supported:
 - `sigmoid`: :math:`y = \frac{1}{1 + exp(-x)}`
 - `tanh`: Hyperbolic tangent, :math:`y = \frac{exp(x) - exp(-x)}{exp(x) + 
exp(-x)}`
 - `softrelu`: Soft ReLU, or SoftPlus, :math:`y = log(1 + exp(x))`
+- `softsign`: :math:`y = \frac{x}{1 + abs(x)}`
 
 )code" ADD_FILELINE)
 .set_attr_parser(ParamParser)
diff --git a/src/operator/operator_tune.cc b/src/operator/operator_tune.cc
index e0f8306..c13f1ac 100644
--- a/src/operator/operator_tune.cc
+++ b/src/operator/operator_tune.cc
@@ -213,6 +213,8 @@ 
IMPLEMENT_UNARY_WORKLOAD_FWD(mxnet::op::mshadow_op::reciprocal);  // NOLINT()
 IMPLEMENT_UNARY_WORKLOAD_BWD(mxnet::op::mshadow_op::reciprocal_grad);  // 
NOLINT()
 IMPLEMENT_UNARY_WORKLOAD_FWD(mxnet::op::mshadow_op::sigmoid);  // NOLINT()
 IMPLEMENT_UNARY_WORKLOAD_BWD(mxnet::op::mshadow_op::sigmoid_grad);  // NOLINT()
+IMPLEMENT_UNARY_WORKLOAD_FWD(mxnet::op::mshadow_op::softsign);  // NOLINT()
+IMPLEMENT_UNARY_WORKLOAD_BWD(mxnet::op::mshadow_op::softsign_grad);  // 
NOLINT()
 IMPLEMENT_UNARY_WORKLOAD_FWD(mxnet::op::mshadow_op::relu);  // NOLINT()
 IMPLEMENT_UNARY_WORKLOAD_BWD(mxnet::op::mshadow_op::relu_grad);  // NOLINT()
 IMPLEMENT_UNARY_WORKLOAD_FWD(mxnet::op::mshadow_op::tanh);  // NOLINT()
diff --git 

[GitHub] Maratyszcza commented on issue #9860: [WIP] CMake NNPack support

[GitBox]

Maratyszcza commented on issue #9860: [WIP] CMake NNPack support
URL: https://github.com/apache/incubator-mxnet/pull/9860#issuecomment-367873330
 
 
   @piiswrong There is no bundled repo. The options are to add NNPACK submodule 
and let it download all deps at configuration time, or add NNPACK and all its 
deps as submodules. Caffe2 started with the first option, but many people 
complained about unexpected downloads at configuration time, and now it 
packages all NNPACK deps as submodules.


This is an automated message from the Apache Git Service.
To respond to the message, please log on GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


With regards,
Apache Git Services

[GitHub] pengzhao-intel commented on issue #9744: Performance regression when OMP_NUM_THREADS environment variable is not set

[GitBox]

pengzhao-intel commented on issue #9744: Performance regression when 
OMP_NUM_THREADS environment variable is not set
URL: 
https://github.com/apache/incubator-mxnet/issues/9744#issuecomment-367873227
 
 
   Thanks, @cjolivier01,  I mean we set this environment in the code as 
OMP_NUM_THREADS.
   
   Actually, this setting is useful when hyperthreading is ON and we only want 
to run each thread in a physical core as below picture shown 
(KMP_AFFINITY=granularity=fine,compact,1,0).
   
   So, this environment will guide how to map thread into the physical core.
   https://software.intel.com/en-us/node/522691
   
   
   ![guid-80fda344-d8c1-4fea-b268-94f685d19c4c-imageid 
0abe3e2e-7a5a-4072-b22f-da7f72eb9a18](https://user-images.githubusercontent.com/22461308/36572333-23cebc60-1876-11e8-9fa8-11004a562ca8.jpg)
   **KMP_AFFINITY=granularity=fine,compact,1,0**
   
   
   


This is an automated message from the Apache Git Service.
To respond to the message, please log on GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


With regards,
Apache Git Services

[GitHub] marcoabreu commented on issue #9798: fix cmake

[GitBox]

marcoabreu commented on issue #9798: fix cmake
URL: https://github.com/apache/incubator-mxnet/pull/9798#issuecomment-367872877
 
 
   Would you mind adding it to the build args in the Jenkinsfile?


This is an automated message from the Apache Git Service.
To respond to the message, please log on GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


With regards,
Apache Git Services

[GitHub] nswamy commented on issue #9858: softsign activation function

[GitBox]

nswamy commented on issue #9858: softsign activation function
URL: https://github.com/apache/incubator-mxnet/pull/9858#issuecomment-367872369
 
 
   I am closing this PR. Customer is ok to apply the patch on v.0.11. 


This is an automated message from the Apache Git Service.
To respond to the message, please log on GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


With regards,
Apache Git Services

[GitHub] nswamy closed pull request #9858: softsign activation function

[GitBox]

nswamy closed pull request #9858: softsign activation function
URL: https://github.com/apache/incubator-mxnet/pull/9858
 
 
   

This is a PR merged from a forked repository.
As GitHub hides the original diff on merge, it is displayed below for
the sake of provenance:

As this is a foreign pull request (from a fork), the diff is supplied
below (as it won't show otherwise due to GitHub magic):

diff --git a/src/operator/contrib/softsign.cc b/src/operator/contrib/softsign.cc
new file mode 100644
index 00..64bdf94d51
--- /dev/null
+++ b/src/operator/contrib/softsign.cc
@@ -0,0 +1,46 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one
+ * or more contributor license agreements.  See the NOTICE file
+ * distributed with this work for additional information
+ * regarding copyright ownership.  The ASF licenses this file
+ * to you under the Apache License, Version 2.0 (the
+ * "License"); you may not use this file except in compliance
+ * with the License.  You may obtain a copy of the License at
+ *
+ *   http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing,
+ * software distributed under the License is distributed on an
+ * "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
+ * KIND, either express or implied.  See the License for the
+ * specific language governing permissions and limitations
+ * under the License.
+ */
+
+/*!
+ * \file softsign.cc
+ * \brief CPU Implementation of softsign function.
+ */
+#include "../tensor/elemwise_unary_op.h"
+#include "../tensor/elemwise_binary_op.h"
+
+namespace mxnet {
+namespace op {
+// softsign
+MXNET_OPERATOR_REGISTER_UNARY(_contrib_softsign)
+  .describe(R"code(Computes softsign of x element-wise.
+
+.. math::
+   y = x / (1 + abs(x))
+
+)code" ADD_FILELINE)
+  .set_attr("FGradient", 
ElemwiseGradUseIn{"_backward_contrib_softsign"})
+  .set_attr("FCompute",
+  UnaryLaunch);
+
+
+MXNET_OPERATOR_REGISTER_BINARY(_backward_contrib_softsign)
+.set_attr("FCompute",
+BinaryLaunch);
+}  // namespace op
+}  // namespace mxnet
diff --git a/src/operator/contrib/softsign.cu b/src/operator/contrib/softsign.cu
new file mode 100644
index 00..7623688475
--- /dev/null
+++ b/src/operator/contrib/softsign.cu
@@ -0,0 +1,38 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one
+ * or more contributor license agreements.  See the NOTICE file
+ * distributed with this work for additional information
+ * regarding copyright ownership.  The ASF licenses this file
+ * to you under the Apache License, Version 2.0 (the
+ * "License"); you may not use this file except in compliance
+ * with the License.  You may obtain a copy of the License at
+ *
+ *   http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing,
+ * software distributed under the License is distributed on an
+ * "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
+ * KIND, either express or implied.  See the License for the
+ * specific language governing permissions and limitations
+ * under the License.
+ */
+
+/*!
+ * \file softsign.cu
+ * \brief CPU Implementation of softsign function.
+ */
+#include "../tensor/elemwise_unary_op.h"
+#include "../tensor/elemwise_binary_op.h"
+
+namespace mxnet {
+namespace op {
+// softsign
+NNVM_REGISTER_OP(_contrib_softsign)
+.set_attr("FCompute", UnaryLaunch);
+
+NNVM_REGISTER_OP(_backward_contrib_softsign)
+.set_attr("FCompute", BinaryLaunch);
+}  // namespace op
+}  // namespace mxnet
+
+
diff --git a/src/operator/mshadow_op.h b/src/operator/mshadow_op.h
index f7815d2f8d..0c4f5bc19b 100644
--- a/src/operator/mshadow_op.h
+++ b/src/operator/mshadow_op.h
@@ -186,6 +186,19 @@ struct softrelu_grad {
 return -DType(expm1f(-a));
   }
 };
+/*! \brief softsign unit */
+struct softsign {
+  template
+  MSHADOW_XINLINE static DType Map(DType a) {
+return DType(a / (DType(1.0f) + fabsf(a)));
+  }
+};
+struct softsign_grad {
+  template
+  MSHADOW_XINLINE static DType Map(DType a) {
+return DType(1.0f / powf((DType(1.0f) + fabsf(a)), 2.0f) );
+  }
+};
 
 struct exp {
   template
diff --git a/src/operator/tensor/elemwise_unary_op.h 
b/src/operator/tensor/elemwise_unary_op.h
index b6994844e0..fc03f08361 100644
--- a/src/operator/tensor/elemwise_unary_op.h
+++ b/src/operator/tensor/elemwise_unary_op.h
@@ -170,6 +170,21 @@ struct relu_grad {
 out[i] = out_grad[i] * DType(in[i] > DType(0.0f) ? DType(1.0f) : 
DType(0.0f));
   }
 };
+/*! \brief softsign unit */
+struct softsign {
+  template
+  MSHADOW_XINLINE static void Map(int i, DType *out,
+  const DType *in) {
+out[i] = DType(DType(in[i]) / (DType(1.0f) + fabsf(in[i])));
+  }
+};
+struct softsign_grad {
+  template
+  MSHADOW_XINLINE static void Map(int i, 

[GitHub] marcoabreu commented on issue #9862: Fix a race condition in converting data layouts in MKLDNN.

[GitBox]

marcoabreu commented on issue #9862: Fix a race condition in converting data 
layouts in MKLDNN.
URL: https://github.com/apache/incubator-mxnet/pull/9862#issuecomment-367872237
 
 
   Would it be possible to create a test that introduces a delay in order to 
force this race condition? 


This is an automated message from the Apache Git Service.
To respond to the message, please log on GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


With regards,
Apache Git Services

[GitHub] yajiedesign commented on issue #9798: fix cmake

[GitBox]

yajiedesign commented on issue #9798: fix cmake
URL: https://github.com/apache/incubator-mxnet/pull/9798#issuecomment-367869677
 
 
   @marcoabreu no,not use in Jenkinsfile.


This is an automated message from the Apache Git Service.
To respond to the message, please log on GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


With regards,
Apache Git Services

[GitHub] zheng-da commented on issue #9862: Fix a race condition in converting data layouts in MKLDNN.

[GitBox]

zheng-da commented on issue #9862: Fix a race condition in converting data 
layouts in MKLDNN.
URL: https://github.com/apache/incubator-mxnet/pull/9862#issuecomment-367869410
 
 
   The reason I disabled the inference tests because I previously thought the 
failure was related to numeric errors and these are invalid tests. Now it's 
clear to me that the failures are caused by race condition. It's very hard to 
reproduce the failure, as you probably have seen. Adding the tests back 
increases the chance of reproducing the failure if the race condition still 
exists. 


This is an automated message from the Apache Git Service.
To respond to the message, please log on GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


With regards,
Apache Git Services

[GitHub] zheng-da commented on issue #9853: Flaky test_operator.test_binary_op

[GitBox]

zheng-da commented on issue #9853: Flaky test_operator.test_binary_op
URL: 
https://github.com/apache/incubator-mxnet/issues/9853#issuecomment-367868599
 
 
   I don't have a clue right now. so far we see failures in random generators 
and binary operators. it's weird why it fails in these simple operators that 
are seemingly irrelevant to MKLDNN operators.


This is an automated message from the Apache Git Service.
To respond to the message, please log on GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


With regards,
Apache Git Services

[GitHub] sampathchanda commented on issue #9100: Python - ImportError: cannot import name 'c_array_buf'

[GitBox]

sampathchanda commented on issue #9100: Python - ImportError: cannot import 
name 'c_array_buf'
URL: 
https://github.com/apache/incubator-mxnet/issues/9100#issuecomment-367866625
 
 
   Am getting the same issue, even when having mxnet installed from 'pip 
install mxnet-cu80==0.11.0'. Any ideas about what could be the fix for that 
case?


This is an automated message from the Apache Git Service.
To respond to the message, please log on GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


With regards,
Apache Git Services

[GitHub] marcoabreu commented on issue #9862: Fix a race condition in converting data layouts in MKLDNN.

[GitBox]

marcoabreu commented on issue #9862: Fix a race condition in converting data 
layouts in MKLDNN.
URL: https://github.com/apache/incubator-mxnet/pull/9862#issuecomment-367866215
 
 
   I'm afraid I still don't really understand. I see that you have re-enabled 
this test, but unfortunately I'm lacking deep technical knowledge of the 
underlying operations, so would you mind explaining me how this re-enabled test 
would have caught this failure?


This is an automated message from the Apache Git Service.
To respond to the message, please log on GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


With regards,
Apache Git Services

[GitHub] cjolivier01 commented on issue #9853: Flaky test_operator.test_binary_op

[GitBox]

cjolivier01 commented on issue #9853: Flaky test_operator.test_binary_op
URL: 
https://github.com/apache/incubator-mxnet/issues/9853#issuecomment-367866040
 
 
   i don?t know what is invoked in the process of calling test_bmod(). could
   be that elemwise_add() isn?t called, or is called before and corrupts
   memory, or maybe has nothing to do with elemwise_add.
   
   however, we seem to have a lot of tests that are suddenly failing... any
   ideas?
   
   On Thu, Feb 22, 2018 at 3:38 PM Da Zheng  wrote:
   
   > @cjolivier01  in both cases (#9853
   >  and #9844
   > ), the tests fail
   > in test_bmod. It shouldn't have invoked elemwise_add
   >
   > ?
   > You are receiving this because you were mentioned.
   >
   >
   > Reply to this email directly, view it on GitHub
   > 
,
   > or mute the thread
   > 

   > .
   >
   


This is an automated message from the Apache Git Service.
To respond to the message, please log on GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


With regards,
Apache Git Services

[GitHub] marcoabreu commented on a change in pull request #9862: Fix a race condition in converting data layouts in MKLDNN.

[GitBox]

marcoabreu commented on a change in pull request #9862: Fix a race condition in 
converting data layouts in MKLDNN.
URL: https://github.com/apache/incubator-mxnet/pull/9862#discussion_r170132857
 
 

 ##
 File path: tests/python/gpu/test_gluon_model_zoo_gpu.py
 ##
 @@ -37,8 +37,7 @@ def download_data():
 return mx.test_utils.download(
 'http://data.mxnet.io/data/val-5k-256.rec', VAL_DATA)
 
-@unittest.skip("test fails intermittently. temporarily disabled.")
-@with_seed()
+@with_seed(1)
 
 Review comment:
   Please remove custom seed


This is an automated message from the Apache Git Service.
To respond to the message, please log on GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


With regards,
Apache Git Services

[GitHub] marcoabreu commented on a change in pull request #9862: Fix a race condition in converting data layouts in MKLDNN.

[GitBox]

marcoabreu commented on a change in pull request #9862: Fix a race condition in 
converting data layouts in MKLDNN.
URL: https://github.com/apache/incubator-mxnet/pull/9862#discussion_r170132926
 
 

 ##
 File path: tests/python/gpu/test_gluon_model_zoo_gpu.py
 ##
 @@ -99,7 +100,7 @@ def get_nn_model(name):
 # Seed 1521019752 produced a failure on the Py2 MKLDNN-GPU CI runner
 # on 2/16/2018 that was not reproducible.  Problem could be timing related or
 # based on non-deterministic algo selection.
-@with_seed()
+@with_seed(1)
 
 Review comment:
   Please remove custom seed


This is an automated message from the Apache Git Service.
To respond to the message, please log on GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


With regards,
Apache Git Services

[GitHub] marcoabreu commented on a change in pull request #9862: Fix a race condition in converting data layouts in MKLDNN.

[GitBox]

marcoabreu commented on a change in pull request #9862: Fix a race condition in 
converting data layouts in MKLDNN.
URL: https://github.com/apache/incubator-mxnet/pull/9862#discussion_r170132782
 
 

 ##
 File path: src/ndarray/ndarray.cc
 ##
 @@ -1017,6 +1017,7 @@ inline void CopyFromToDnsImpl(const NDArray& from, const 
NDArray& to, RunContext
 // with Copy().
 NDArray tmp_from = from;
 if (tmp_from.IsMKLDNNData()) {
+  // TODO(zhengda) tmp_from should be cached.
 
 Review comment:
   In think it's best practice and that we should lead by example :) In the 
last months I haven't seen any TODO (besides the ones in your MKLDNN PR) 
checked into MXNet. 
   
   We got rid of the script to automatically close issues. For now, GitHub 
issues are the right place to track these things.


This is an automated message from the Apache Git Service.
To respond to the message, please log on GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


With regards,
Apache Git Services

[GitHub] marcoabreu commented on issue #9862: Fix a race condition in converting data layouts in MKLDNN.

[GitBox]

marcoabreu commented on issue #9862: Fix a race condition in converting data 
layouts in MKLDNN.
URL: https://github.com/apache/incubator-mxnet/pull/9862#issuecomment-367865250
 
 
   We don't need fixed seeds since #9791 has been merged. Just make sure that 
the with_seed decorator is properly applied.


This is an automated message from the Apache Git Service.
To respond to the message, please log on GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


With regards,
Apache Git Services

[GitHub] zheng-da commented on a change in pull request #9862: Fix a race condition in converting data layouts in MKLDNN.

[GitBox]

zheng-da commented on a change in pull request #9862: Fix a race condition in 
converting data layouts in MKLDNN.
URL: https://github.com/apache/incubator-mxnet/pull/9862#discussion_r170130612
 
 

 ##
 File path: src/ndarray/ndarray.cc
 ##
 @@ -1017,6 +1017,7 @@ inline void CopyFromToDnsImpl(const NDArray& from, const 
NDArray& to, RunContext
 // with Copy().
 NDArray tmp_from = from;
 if (tmp_from.IsMKLDNNData()) {
+  // TODO(zhengda) tmp_from should be cached.
 
 Review comment:
   The reason I ask is that I saw TODO in many places in the code. Are all 
TODOs tracked in github issues? Or is this the new practice we want to push? 
And is github issue the best place for us to keep track of these TODOs? I 
remember Sheng has a script that closes issues if they aren't inactive for 30 
days or so.


This is an automated message from the Apache Git Service.
To respond to the message, please log on GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


With regards,
Apache Git Services

[GitHub] zheng-da commented on issue #9862: Fix a race condition in converting data layouts in MKLDNN.

[GitBox]

zheng-da commented on issue #9862: Fix a race condition in converting data 
layouts in MKLDNN.
URL: https://github.com/apache/incubator-mxnet/pull/9862#issuecomment-367862145
 
 
   @marcoabreu enabling the tests can catch the error more easily.


This is an automated message from the Apache Git Service.
To respond to the message, please log on GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


With regards,
Apache Git Services

[GitHub] zheng-da commented on issue #9862: Fix a race condition in converting data layouts in MKLDNN.

[GitBox]

zheng-da commented on issue #9862: Fix a race condition in converting data 
layouts in MKLDNN.
URL: https://github.com/apache/incubator-mxnet/pull/9862#issuecomment-367861997
 
 
   @cjolivier01 The seed is set so we know what is the expected result. It's 
easier to tell whether CPU or GPU compute wrong results.


This is an automated message from the Apache Git Service.
To respond to the message, please log on GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


With regards,
Apache Git Services

[GitHub] zheng-da commented on a change in pull request #9862: Fix a race condition in converting data layouts in MKLDNN.

[GitBox]

zheng-da commented on a change in pull request #9862: Fix a race condition in 
converting data layouts in MKLDNN.
URL: https://github.com/apache/incubator-mxnet/pull/9862#discussion_r170129335
 
 

 ##
 File path: src/operator/nn/mkldnn/mkldnn_base.cc
 ##
 @@ -270,9 +270,24 @@ void FallBackCompute(FCompute fn, const nnvm::NodeAttrs 
,
  const std::vector ,
  const std::vector ) {
   std::vector in_blobs(inputs.size());
+  std::vector in_bufs;
   for (size_t i = 0; i < in_blobs.size(); i++) {
+// If the input data isn't stored in the default format, we shouldn't
+// call data() directly, which will change the layout of the NDArray.
+// Instead, we should save the converted data in another NDArray.
+// TODO(zhengda) we should use temp space to save the converted data.
+if (inputs[i].IsDefaultData()) {
   in_blobs[i] = inputs[i].data();
+} else {
+  in_bufs.emplace_back(inputs[i].shape(), inputs[i].ctx(),
 
 Review comment:
   I saw reserve() being used in many places in MXNet. I'm always wondering how 
much benefit it brings. Or is this a convention?


This is an automated message from the Apache Git Service.
To respond to the message, please log on GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


With regards,
Apache Git Services

[GitHub] zheng-da commented on issue #9853: Flaky test_operator.test_binary_op

[GitBox]

zheng-da commented on issue #9853: Flaky test_operator.test_binary_op
URL: 
https://github.com/apache/incubator-mxnet/issues/9853#issuecomment-367860345
 
 
   @cjolivier01 in both cases 
(https://github.com/apache/incubator-mxnet/issues/9853 and 
https://github.com/apache/incubator-mxnet/issues/9844), the tests fail in 
test_bmod. It shouldn't have invoked elemwise_add 


This is an automated message from the Apache Git Service.
To respond to the message, please log on GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


With regards,
Apache Git Services

[GitHub] marcoabreu commented on issue #9853: Flaky test_operator.test_binary_op

[GitBox]

marcoabreu commented on issue #9853: Flaky test_operator.test_binary_op
URL: 
https://github.com/apache/incubator-mxnet/issues/9853#issuecomment-367860051
 
 
   Well even if it was changed to use MKL, this would not apply here since 
we're running on OpenBLAS, right?


This is an automated message from the Apache Git Service.
To respond to the message, please log on GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


With regards,
Apache Git Services

[GitHub] nswamy commented on issue #9858: softsign activation function

[GitBox]

nswamy commented on issue #9858: softsign activation function
URL: https://github.com/apache/incubator-mxnet/pull/9858#issuecomment-367859595
 
 
   I am ok to put it to all the releases after 0.11.0 upto master, but I don't 
think it has to be done right away. 


This is an automated message from the Apache Git Service.
To respond to the message, please log on GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


With regards,
Apache Git Services

[GitHub] marcoabreu commented on issue #9853: Flaky test_operator.test_binary_op

[GitBox]

marcoabreu commented on issue #9853: Flaky test_operator.test_binary_op
URL: 
https://github.com/apache/incubator-mxnet/issues/9853#issuecomment-367859620
 
 
   I don't think we have the tools to measure performance on that scale yet. As 
far as I know, this is in the works. Since there's still some time until 1.2, 
we can definitely gather these numbers.


This is an automated message from the Apache Git Service.
To respond to the message, please log on GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


With regards,
Apache Git Services

[GitHub] marcoabreu commented on issue #9843: Flaky test_random.test_exponential_generator @ Python2: MKLDNN-CPU

[GitBox]

marcoabreu commented on issue #9843: Flaky 
test_random.test_exponential_generator @ Python2: MKLDNN-CPU
URL: 
https://github.com/apache/incubator-mxnet/issues/9843#issuecomment-367858507
 
 
   That might fix the test, but could we risk masking a problem that way? I'm 
afraid I'm unfamiliar with the topic, so would you mind explaining how an 
actual failure would look like and how we could select the best threshold? 


This is an automated message from the Apache Git Service.
To respond to the message, please log on GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


With regards,
Apache Git Services

[GitHub] cjolivier01 commented on issue #9858: softsign activation function

[GitBox]

cjolivier01 commented on issue #9858: softsign activation function
URL: https://github.com/apache/incubator-mxnet/pull/9858#issuecomment-367858332
 
 
   +1 would have to go into all releases after 0.11


This is an automated message from the Apache Git Service.
To respond to the message, please log on GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


With regards,
Apache Git Services

[GitHub] marcoabreu commented on issue #9856: Flaky test_random.test_gamma_generator

[GitBox]

marcoabreu commented on issue #9856: Flaky test_random.test_gamma_generator
URL: 
https://github.com/apache/incubator-mxnet/issues/9856#issuecomment-367857691
 
 
   No, it's using OpenBLAS.


This is an automated message from the Apache Git Service.
To respond to the message, please log on GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


With regards,
Apache Git Services

[GitHub] piiswrong commented on issue #9858: softsign activation function

[GitBox]

piiswrong commented on issue #9858: softsign activation function
URL: https://github.com/apache/incubator-mxnet/pull/9858#issuecomment-367857655
 
 
   We can't just put stuff into arbitrary previous versions and then have it 
disappear later then reappear.
   
   I would just make a fork for this customer.


This is an automated message from the Apache Git Service.
To respond to the message, please log on GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


With regards,
Apache Git Services

[GitHub] marcoabreu commented on issue #9858: softsign activation function

[GitBox]

marcoabreu commented on issue #9858: softsign activation function
URL: https://github.com/apache/incubator-mxnet/pull/9858#issuecomment-367857486
 
 
   Contrib also sounds good to me. Are we going to make a formal release or 
will that customer just checkout the branch head and compile themselves?


This is an automated message from the Apache Git Service.
To respond to the message, please log on GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


With regards,
Apache Git Services

[GitHub] nswamy commented on issue #9858: softsign activation function

[GitBox]

nswamy commented on issue #9858: softsign activation function
URL: https://github.com/apache/incubator-mxnet/pull/9858#issuecomment-367855882
 
 
   Ok, I registered the symbol under contrib.


This is an automated message from the Apache Git Service.
To respond to the message, please log on GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


With regards,
Apache Git Services

[GitHub] marcoabreu commented on issue #9864: Flaky hanging test_operator.test_laop_3 @ Python3: MKLDNN-CPU

[GitBox]

marcoabreu commented on issue #9864: Flaky hanging test_operator.test_laop_3 @ 
Python3: MKLDNN-CPU
URL: 
https://github.com/apache/incubator-mxnet/issues/9864#issuecomment-367853511
 
 
   @asmushetzel would you like to investigate?


This is an automated message from the Apache Git Service.
To respond to the message, please log on GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


With regards,
Apache Git Services

[GitHub] marcoabreu opened a new issue #9864: Flaky hanging @ Python3: MKLDNN-CPU

[GitBox]

marcoabreu opened a new issue #9864: Flaky hanging @ Python3: MKLDNN-CPU
URL: https://github.com/apache/incubator-mxnet/issues/9864
 
 
   Test times out due to hang.
   
   
http://jenkins.mxnet-ci.amazon-ml.com/blue/organizations/jenkins/incubator-mxnet/detail/PR-9848/1/pipeline
   
   ```
   test_operator.test_laop_3 ... [INFO] Setting test np/mx/python random seeds, 
use MXNET_TEST_SEED=1896893923 to reproduce.
   
   Sending interrupt signal to process
   
   Terminated
   
   script returned exit code 143
   ```
   
   


This is an automated message from the Apache Git Service.
To respond to the message, please log on GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


With regards,
Apache Git Services

[GitHub] cjolivier01 commented on issue #9848: add infer_type for regression ops, fix issues#9847

[GitBox]

cjolivier01 commented on issue #9848: add infer_type for regression ops, fix 
issues#9847
URL: https://github.com/apache/incubator-mxnet/pull/9848#issuecomment-367853082
 
 
   Please feel free to add infer type for the other ones if you like (not 
required for this PR, however).


This is an automated message from the Apache Git Service.
To respond to the message, please log on GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


With regards,
Apache Git Services

[GitHub] marcoabreu commented on issue #9848: add infer_type for regression ops, fix issues#9847

[GitBox]

marcoabreu commented on issue #9848: add infer_type for regression ops, fix 
issues#9847
URL: https://github.com/apache/incubator-mxnet/pull/9848#issuecomment-367852882
 
 
   Very nice catch, @ZiyueHuang! Sounds like a good idea.


This is an automated message from the Apache Git Service.
To respond to the message, please log on GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


With regards,
Apache Git Services

[GitHub] cjolivier01 commented on issue #9848: add infer_type for regression ops, fix issues#9847

[GitBox]

cjolivier01 commented on issue #9848: add infer_type for regression ops, fix 
issues#9847
URL: https://github.com/apache/incubator-mxnet/pull/9848#issuecomment-367852782
 
 
   Makes sense.


This is an automated message from the Apache Git Service.
To respond to the message, please log on GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


With regards,
Apache Git Services

[GitHub] marcoabreu commented on issue #9862: Fix a race condition in converting data layouts in MKLDNN.

[GitBox]

marcoabreu commented on issue #9862: Fix a race condition in converting data 
layouts in MKLDNN.
URL: https://github.com/apache/incubator-mxnet/pull/9862#issuecomment-367850610
 
 
   But would this re-enabled test have caught this error? Please explain how 
we're preventing this problem from reoccurring.


This is an automated message from the Apache Git Service.
To respond to the message, please log on GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


With regards,
Apache Git Services

[GitHub] cjolivier01 commented on issue #9853: Flaky test_operator.test_binary_op

[GitBox]

cjolivier01 commented on issue #9853: Flaky test_operator.test_binary_op
URL: 
https://github.com/apache/incubator-mxnet/issues/9853#issuecomment-367850176
 
 
   Wasn't elemwise_add changed to use mkl?
   btw, was it verified that mkl is faster for all shapes and types?  I saw it 
allocates memory, which seems like it might be slow.


This is an automated message from the Apache Git Service.
To respond to the message, please log on GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


With regards,
Apache Git Services

[GitHub] marcoabreu commented on a change in pull request #9862: Fix a race condition in converting data layouts in MKLDNN.

[GitBox]

marcoabreu commented on a change in pull request #9862: Fix a race condition in 
converting data layouts in MKLDNN.
URL: https://github.com/apache/incubator-mxnet/pull/9862#discussion_r170119050
 
 

 ##
 File path: src/ndarray/ndarray.cc
 ##
 @@ -1017,6 +1017,7 @@ inline void CopyFromToDnsImpl(const NDArray& from, const 
NDArray& to, RunContext
 // with Copy().
 NDArray tmp_from = from;
 if (tmp_from.IsMKLDNNData()) {
+  // TODO(zhengda) tmp_from should be cached.
 
 Review comment:
   Yes. Ideally, there should be no TODOs at all. But in cases like this it's 
acceptable to have them, but they must be tracked as otherwise we lose overview.


This is an automated message from the Apache Git Service.
To respond to the message, please log on GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


With regards,
Apache Git Services

[GitHub] sxjscience commented on issue #9843: Flaky test_random.test_exponential_generator @ Python2: MKLDNN-CPU

[GitBox]

sxjscience commented on issue #9843: Flaky 
test_random.test_exponential_generator @ Python2: MKLDNN-CPU
URL: 
https://github.com/apache/incubator-mxnet/issues/9843#issuecomment-367849367
 
 
   As shown in the log, the chi-square value is above the 0.05 threshold 
(indicates the check has passed) once in the 5 runs, i.e, 20% success rate. 
Currently we force the success rate to be 25% 
(https://github.com/apache/incubator-mxnet/blob/master/python/mxnet/test_utils.py#L1848),
 we could lower it to 10% to fix the test.


This is an automated message from the Apache Git Service.
To respond to the message, please log on GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


With regards,
Apache Git Services

[GitHub] sxjscience commented on issue #9843: Flaky test_random.test_exponential_generator @ Python2: MKLDNN-CPU

[GitBox]

sxjscience commented on issue #9843: Flaky 
test_random.test_exponential_generator @ Python2: MKLDNN-CPU
URL: 
https://github.com/apache/incubator-mxnet/issues/9843#issuecomment-367849367
 
 
   As shown in the log, the chi-square value is above the 0.05 threshold 
(indicates the check has passed) once in the 5 runs, i.e, 20% success rate. 
Currently we force the success rate to be 25%, we could lower it to 10% to fix 
the test.


This is an automated message from the Apache Git Service.
To respond to the message, please log on GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


With regards,
Apache Git Services

[GitHub] marcoabreu opened a new pull request #9863: Fix broken website build and update version number

[GitBox]

marcoabreu opened a new pull request #9863: Fix broken website build and update 
version number
URL: https://github.com/apache/incubator-mxnet/pull/9863
 
 
   Update website build script to point to the right file
   Add 1.1.0 to build_all_versions.sh
   Made build scripts fail fast with set -xe


This is an automated message from the Apache Git Service.
To respond to the message, please log on GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


With regards,
Apache Git Services

[GitHub] cjolivier01 commented on a change in pull request #9862: Fix a race condition in converting data layouts in MKLDNN.

[GitBox]

cjolivier01 commented on a change in pull request #9862: Fix a race condition 
in converting data layouts in MKLDNN.
URL: https://github.com/apache/incubator-mxnet/pull/9862#discussion_r170117620
 
 

 ##
 File path: tests/python/gpu/test_gluon_model_zoo_gpu.py
 ##
 @@ -37,11 +37,11 @@ def download_data():
 return mx.test_utils.download(
 'http://data.mxnet.io/data/val-5k-256.rec', VAL_DATA)
 
-@unittest.skip("test fails intermittently. temporarily disabled.")
 @with_seed()
 def test_inference():
 all_models = ['resnet50_v1', 'vgg19_bn', 'alexnet', #'inceptionv3',
   'densenet201', 'squeezenet1.0', 'mobilenet0.25']
+mx.random.seed(2)
 
 Review comment:
   Why would you want to set it the same for each run?


This is an automated message from the Apache Git Service.
To respond to the message, please log on GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


With regards,
Apache Git Services

[GitHub] cjolivier01 commented on a change in pull request #9862: Fix a race condition in converting data layouts in MKLDNN.

[GitBox]

cjolivier01 commented on a change in pull request #9862: Fix a race condition 
in converting data layouts in MKLDNN.
URL: https://github.com/apache/incubator-mxnet/pull/9862#discussion_r170117170
 
 

 ##
 File path: src/operator/nn/mkldnn/mkldnn_base.cc
 ##
 @@ -270,9 +270,24 @@ void FallBackCompute(FCompute fn, const nnvm::NodeAttrs 
,
  const std::vector ,
  const std::vector ) {
   std::vector in_blobs(inputs.size());
+  std::vector in_bufs;
   for (size_t i = 0; i < in_blobs.size(); i++) {
+// If the input data isn't stored in the default format, we shouldn't
+// call data() directly, which will change the layout of the NDArray.
+// Instead, we should save the converted data in another NDArray.
+// TODO(zhengda) we should use temp space to save the converted data.
+if (inputs[i].IsDefaultData()) {
   in_blobs[i] = inputs[i].data();
+} else {
+  in_bufs.emplace_back(inputs[i].shape(), inputs[i].ctx(),
+   false, inputs[i].dtype());
+  auto mem = inputs[i].GetMKLDNNData();
 
 Review comment:
   nit: auto should be used for obvious types (ie auto foo = static_cast(bar)) or stuff like iterators which are super-long.  This is just a simnple 
var.  For readability here, it's be awesome to see what this type actually is.


This is an automated message from the Apache Git Service.
To respond to the message, please log on GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


With regards,
Apache Git Services

[GitHub] cjolivier01 commented on a change in pull request #9862: Fix a race condition in converting data layouts in MKLDNN.

[GitBox]

cjolivier01 commented on a change in pull request #9862: Fix a race condition 
in converting data layouts in MKLDNN.
URL: https://github.com/apache/incubator-mxnet/pull/9862#discussion_r170116831
 
 

 ##
 File path: src/operator/nn/mkldnn/mkldnn_base.cc
 ##
 @@ -270,9 +270,24 @@ void FallBackCompute(FCompute fn, const nnvm::NodeAttrs 
,
  const std::vector ,
  const std::vector ) {
   std::vector in_blobs(inputs.size());
+  std::vector in_bufs;
   for (size_t i = 0; i < in_blobs.size(); i++) {
+// If the input data isn't stored in the default format, we shouldn't
+// call data() directly, which will change the layout of the NDArray.
+// Instead, we should save the converted data in another NDArray.
+// TODO(zhengda) we should use temp space to save the converted data.
+if (inputs[i].IsDefaultData()) {
   in_blobs[i] = inputs[i].data();
+} else {
+  in_bufs.emplace_back(inputs[i].shape(), inputs[i].ctx(),
 
 Review comment:
   nit: it would be nice to call in_bufs.reserve(in_blobs.size()) on this first 
pass (ie if in_bufs.empty())
   If there is often more than one in_bufs to be inserted


This is an automated message from the Apache Git Service.
To respond to the message, please log on GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


With regards,
Apache Git Services

[GitHub] zheng-da commented on issue #9853: Flaky test_operator.test_binary_op

[GitBox]

zheng-da commented on issue #9853: Flaky test_operator.test_binary_op
URL: 
https://github.com/apache/incubator-mxnet/issues/9853#issuecomment-367846746
 
 
   It seems https://github.com/apache/incubator-mxnet/issues/9853 fails in the 
same test. It's strange how this fails.


This is an automated message from the Apache Git Service.
To respond to the message, please log on GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


With regards,
Apache Git Services

[GitHub] zheng-da commented on issue #9862: Fix a race condition in converting data layouts in MKLDNN.

[GitBox]

zheng-da commented on issue #9862: Fix a race condition in converting data 
layouts in MKLDNN.
URL: https://github.com/apache/incubator-mxnet/pull/9862#issuecomment-367845835
 
 
   @marcoabreu previously, I disabled the inference test. Now I enabled all 
tests. I also added some prints to clearly see whether CPU or GPU generates 
wrong results.


This is an automated message from the Apache Git Service.
To respond to the message, please log on GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


With regards,
Apache Git Services

[GitHub] zheng-da commented on a change in pull request #9862: Fix a race condition in converting data layouts in MKLDNN.

[GitBox]

zheng-da commented on a change in pull request #9862: Fix a race condition in 
converting data layouts in MKLDNN.
URL: https://github.com/apache/incubator-mxnet/pull/9862#discussion_r170114168
 
 

 ##
 File path: tests/python/gpu/test_gluon_model_zoo_gpu.py
 ##
 @@ -37,11 +37,11 @@ def download_data():
 return mx.test_utils.download(
 'http://data.mxnet.io/data/val-5k-256.rec', VAL_DATA)
 
-@unittest.skip("test fails intermittently. temporarily disabled.")
 @with_seed()
 def test_inference():
 all_models = ['resnet50_v1', 'vgg19_bn', 'alexnet', #'inceptionv3',
   'densenet201', 'squeezenet1.0', 'mobilenet0.25']
+mx.random.seed(2)
 
 Review comment:
   Does with_seed() set the same seed for each run?


This is an automated message from the Apache Git Service.
To respond to the message, please log on GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


With regards,
Apache Git Services

[GitHub] zheng-da commented on a change in pull request #9862: Fix a race condition in converting data layouts in MKLDNN.

[GitBox]

zheng-da commented on a change in pull request #9862: Fix a race condition in 
converting data layouts in MKLDNN.
URL: https://github.com/apache/incubator-mxnet/pull/9862#discussion_r170114324
 
 

 ##
 File path: src/ndarray/ndarray.cc
 ##
 @@ -1017,6 +1017,7 @@ inline void CopyFromToDnsImpl(const NDArray& from, const 
NDArray& to, RunContext
 // with Copy().
 NDArray tmp_from = from;
 if (tmp_from.IsMKLDNNData()) {
+  // TODO(zhengda) tmp_from should be cached.
 
 Review comment:
   Create a github issue for this?


This is an automated message from the Apache Git Service.
To respond to the message, please log on GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


With regards,
Apache Git Services

[incubator-mxnet] branch master updated: Add CuDNN (v6) support to Jetson builds (#9861)

[marcoabreu]

This is an automated email from the ASF dual-hosted git repository.

marcoabreu pushed a commit to branch master
in repository https://gitbox.apache.org/repos/asf/incubator-mxnet.git


The following commit(s) were added to refs/heads/master by this push:
 new 18816a5  Add CuDNN (v6) support to Jetson builds (#9861)
18816a5 is described below

commit 18816a57a00c94f4db604937e444a145f90aab10
Author: Kellen Sunderland 
AuthorDate: Thu Feb 22 23:28:36 2018 +0100

Add CuDNN (v6) support to Jetson builds (#9861)
---
 docker_multiarch/Dockerfile.build.jetson | 26 +-
 1 file changed, 13 insertions(+), 13 deletions(-)

diff --git a/docker_multiarch/Dockerfile.build.jetson 
b/docker_multiarch/Dockerfile.build.jetson
index 83f8be7..8a60fae 100644
--- a/docker_multiarch/Dockerfile.build.jetson
+++ b/docker_multiarch/Dockerfile.build.jetson
@@ -1,7 +1,7 @@
 # -*- mode: dockerfile -*-
 # dockerfile to build libmxnet.so, and a python wheel for the Jetson TX1/TX2
 
-FROM nvidia/cuda:8.0-cudnn5-devel as cudabuilder
+FROM nvidia/cuda:8.0-cudnn6-devel as cudabuilder
 
 FROM dockcross/linux-arm64
 
@@ -21,13 +21,7 @@ ADD 
https://api.github.com/repos/xianyi/OpenBLAS/git/refs/heads/master /tmp/open
 RUN git clone https://github.com/xianyi/OpenBLAS.git && \
 cd OpenBLAS && \
 make -j$(nproc) TARGET=ARMV8 && \
-make install && \
-ln -s /opt/OpenBLAS/lib/libopenblas.so /usr/lib/libopenblas.so && \
-ln -s /opt/OpenBLAS/lib/libopenblas.a /usr/lib/libopenblas.a && \
-ln -s /opt/OpenBLAS/lib/libopenblas.a /usr/lib/liblapack.a
-
-ENV LD_LIBRARY_PATH $LD_LIBRARY_PATH:/opt/OpenBLAS/lib
-ENV CPLUS_INCLUDE_PATH /opt/OpenBLAS/include
+PREFIX=/usr make install
 
 # Setup CUDA build env (including configuring and copying nvcc)
 COPY --from=cudabuilder /usr/local/cuda /usr/local/cuda
@@ -36,10 +30,16 @@ ENV TARGET_ARCH aarch64
 ENV TARGET_OS linux
 
 # Install ARM depedencies based on Jetpack 3.1
-RUN wget 
http://developer.download.nvidia.com/devzone/devcenter/mobile/jetpack_l4t/013/linux-x64/cuda-repo-l4t-8-0-local_8.0.84-1_arm64.deb
 && \
-wget 
http://developer.download.nvidia.com/devzone/devcenter/mobile/jetpack_l4t/013/linux-x64/libcudnn6_6.0.21-1+cuda8.0_arm64.deb
 && \
-dpkg -i cuda-repo-l4t-8-0-local_8.0.84-1_arm64.deb && \
-dpkg -i libcudnn6_6.0.21-1+cuda8.0_arm64.deb && \
+RUN 
JETPACK_DOWNLOAD_PREFIX=http://developer.download.nvidia.com/devzone/devcenter/mobile/jetpack_l4t/013/linux-x64
 && \
+ARM_CUDA_INSTALLER_PACKAGE=cuda-repo-l4t-8-0-local_8.0.84-1_arm64.deb && \
+ARM_CUDNN_INSTALLER_PACKAGE=libcudnn6_6.0.21-1+cuda8.0_arm64.deb && \
+ARM_CUDNN_DEV_INSTALLER_PACKAGE=libcudnn6-dev_6.0.21-1+cuda8.0_arm64.deb 
&& \
+wget $JETPACK_DOWNLOAD_PREFIX/$ARM_CUDA_INSTALLER_PACKAGE && \
+wget $JETPACK_DOWNLOAD_PREFIX/$ARM_CUDNN_INSTALLER_PACKAGE && \
+wget $JETPACK_DOWNLOAD_PREFIX/$ARM_CUDNN_DEV_INSTALLER_PACKAGE && \
+dpkg -i $ARM_CUDA_INSTALLER_PACKAGE && \
+dpkg -i $ARM_CUDNN_INSTALLER_PACKAGE && \
+dpkg -i $ARM_CUDNN_DEV_INSTALLER_PACKAGE && \
 apt update -y && \
 apt install cuda-cudart-cross-aarch64-8-0 cuda-cublas-cross-aarch64-8-0 \
 cuda-nvml-cross-aarch64-8-0 cuda-nvrtc-cross-aarch64-8-0 
cuda-cufft-cross-aarch64-8-0 \
@@ -48,7 +48,7 @@ RUN wget 
http://developer.download.nvidia.com/devzone/devcenter/mobile/jetpack_l
 cp /usr/local/cuda-8.0/targets/aarch64-linux/lib/*.so 
/usr/local/cuda/lib64/ && \
 cp /usr/local/cuda-8.0/targets/aarch64-linux/lib/stubs/*.so 
/usr/local/cuda/lib64/stubs/ && \
 cp -r /usr/local/cuda-8.0/targets/aarch64-linux/include/ 
/usr/local/cuda/include/ && \
-rm cuda-repo-l4t-8-0-local_8.0.84-1_arm64.deb && rm 
libcudnn6_6.0.21-1+cuda8.0_arm64.deb
+rm $ARM_CUDA_INSTALLER_PACKAGE $ARM_CUDNN_INSTALLER_PACKAGE 
$ARM_CUDNN_DEV_INSTALLER_PACKAGE
 
 # Build MXNet
 ADD mxnet mxnet

-- 
To stop receiving notification emails like this one, please contact
marcoab...@apache.org.

[GitHub] marcoabreu closed pull request #9861: Add CuDNN (v6) support to Jetson builds

[GitBox]

marcoabreu closed pull request #9861: Add CuDNN (v6) support to Jetson builds
URL: https://github.com/apache/incubator-mxnet/pull/9861
 
 
   

This is a PR merged from a forked repository.
As GitHub hides the original diff on merge, it is displayed below for
the sake of provenance:

As this is a foreign pull request (from a fork), the diff is supplied
below (as it won't show otherwise due to GitHub magic):

diff --git a/docker_multiarch/Dockerfile.build.jetson 
b/docker_multiarch/Dockerfile.build.jetson
index 83f8be74fa..8a60faef90 100644
--- a/docker_multiarch/Dockerfile.build.jetson
+++ b/docker_multiarch/Dockerfile.build.jetson
@@ -1,7 +1,7 @@
 # -*- mode: dockerfile -*-
 # dockerfile to build libmxnet.so, and a python wheel for the Jetson TX1/TX2
 
-FROM nvidia/cuda:8.0-cudnn5-devel as cudabuilder
+FROM nvidia/cuda:8.0-cudnn6-devel as cudabuilder
 
 FROM dockcross/linux-arm64
 
@@ -21,13 +21,7 @@ ADD 
https://api.github.com/repos/xianyi/OpenBLAS/git/refs/heads/master /tmp/open
 RUN git clone https://github.com/xianyi/OpenBLAS.git && \
 cd OpenBLAS && \
 make -j$(nproc) TARGET=ARMV8 && \
-make install && \
-ln -s /opt/OpenBLAS/lib/libopenblas.so /usr/lib/libopenblas.so && \
-ln -s /opt/OpenBLAS/lib/libopenblas.a /usr/lib/libopenblas.a && \
-ln -s /opt/OpenBLAS/lib/libopenblas.a /usr/lib/liblapack.a
-
-ENV LD_LIBRARY_PATH $LD_LIBRARY_PATH:/opt/OpenBLAS/lib
-ENV CPLUS_INCLUDE_PATH /opt/OpenBLAS/include
+PREFIX=/usr make install
 
 # Setup CUDA build env (including configuring and copying nvcc)
 COPY --from=cudabuilder /usr/local/cuda /usr/local/cuda
@@ -36,10 +30,16 @@ ENV TARGET_ARCH aarch64
 ENV TARGET_OS linux
 
 # Install ARM depedencies based on Jetpack 3.1
-RUN wget 
http://developer.download.nvidia.com/devzone/devcenter/mobile/jetpack_l4t/013/linux-x64/cuda-repo-l4t-8-0-local_8.0.84-1_arm64.deb
 && \
-wget 
http://developer.download.nvidia.com/devzone/devcenter/mobile/jetpack_l4t/013/linux-x64/libcudnn6_6.0.21-1+cuda8.0_arm64.deb
 && \
-dpkg -i cuda-repo-l4t-8-0-local_8.0.84-1_arm64.deb && \
-dpkg -i libcudnn6_6.0.21-1+cuda8.0_arm64.deb && \
+RUN 
JETPACK_DOWNLOAD_PREFIX=http://developer.download.nvidia.com/devzone/devcenter/mobile/jetpack_l4t/013/linux-x64
 && \
+ARM_CUDA_INSTALLER_PACKAGE=cuda-repo-l4t-8-0-local_8.0.84-1_arm64.deb && \
+ARM_CUDNN_INSTALLER_PACKAGE=libcudnn6_6.0.21-1+cuda8.0_arm64.deb && \
+ARM_CUDNN_DEV_INSTALLER_PACKAGE=libcudnn6-dev_6.0.21-1+cuda8.0_arm64.deb 
&& \
+wget $JETPACK_DOWNLOAD_PREFIX/$ARM_CUDA_INSTALLER_PACKAGE && \
+wget $JETPACK_DOWNLOAD_PREFIX/$ARM_CUDNN_INSTALLER_PACKAGE && \
+wget $JETPACK_DOWNLOAD_PREFIX/$ARM_CUDNN_DEV_INSTALLER_PACKAGE && \
+dpkg -i $ARM_CUDA_INSTALLER_PACKAGE && \
+dpkg -i $ARM_CUDNN_INSTALLER_PACKAGE && \
+dpkg -i $ARM_CUDNN_DEV_INSTALLER_PACKAGE && \
 apt update -y && \
 apt install cuda-cudart-cross-aarch64-8-0 cuda-cublas-cross-aarch64-8-0 \
 cuda-nvml-cross-aarch64-8-0 cuda-nvrtc-cross-aarch64-8-0 
cuda-cufft-cross-aarch64-8-0 \
@@ -48,7 +48,7 @@ RUN wget 
http://developer.download.nvidia.com/devzone/devcenter/mobile/jetpack_l
 cp /usr/local/cuda-8.0/targets/aarch64-linux/lib/*.so 
/usr/local/cuda/lib64/ && \
 cp /usr/local/cuda-8.0/targets/aarch64-linux/lib/stubs/*.so 
/usr/local/cuda/lib64/stubs/ && \
 cp -r /usr/local/cuda-8.0/targets/aarch64-linux/include/ 
/usr/local/cuda/include/ && \
-rm cuda-repo-l4t-8-0-local_8.0.84-1_arm64.deb && rm 
libcudnn6_6.0.21-1+cuda8.0_arm64.deb
+rm $ARM_CUDA_INSTALLER_PACKAGE $ARM_CUDNN_INSTALLER_PACKAGE 
$ARM_CUDNN_DEV_INSTALLER_PACKAGE
 
 # Build MXNet
 ADD mxnet mxnet


 


This is an automated message from the Apache Git Service.
To respond to the message, please log on GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


With regards,
Apache Git Services

[GitHub] marcoabreu commented on a change in pull request #9862: Fix a race condition in converting data layouts in MKLDNN.

[GitBox]

marcoabreu commented on a change in pull request #9862: Fix a race condition in 
converting data layouts in MKLDNN.
URL: https://github.com/apache/incubator-mxnet/pull/9862#discussion_r170113340
 
 

 ##
 File path: tests/python/gpu/test_gluon_model_zoo_gpu.py
 ##
 @@ -37,11 +37,11 @@ def download_data():
 return mx.test_utils.download(
 'http://data.mxnet.io/data/val-5k-256.rec', VAL_DATA)
 
-@unittest.skip("test fails intermittently. temporarily disabled.")
 @with_seed()
 def test_inference():
 all_models = ['resnet50_v1', 'vgg19_bn', 'alexnet', #'inceptionv3',
   'densenet201', 'squeezenet1.0', 'mobilenet0.25']
+mx.random.seed(2)
 
 Review comment:
   please don't use mx.random.seed due to the introduction of the @with_seed() 
decorator


This is an automated message from the Apache Git Service.
To respond to the message, please log on GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


With regards,
Apache Git Services

[GitHub] marcoabreu commented on a change in pull request #9862: Fix a race condition in converting data layouts in MKLDNN.

[GitBox]

marcoabreu commented on a change in pull request #9862: Fix a race condition in 
converting data layouts in MKLDNN.
URL: https://github.com/apache/incubator-mxnet/pull/9862#discussion_r170113089
 
 

 ##
 File path: src/ndarray/ndarray.cc
 ##
 @@ -1017,6 +1017,7 @@ inline void CopyFromToDnsImpl(const NDArray& from, const 
NDArray& to, RunContext
 // with Copy().
 NDArray tmp_from = from;
 if (tmp_from.IsMKLDNNData()) {
+  // TODO(zhengda) tmp_from should be cached.
 
 Review comment:
   Please create an issue to track this


This is an automated message from the Apache Git Service.
To respond to the message, please log on GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


With regards,
Apache Git Services

[GitHub] marcoabreu commented on a change in pull request #9862: Fix a race condition in converting data layouts in MKLDNN.

[GitBox]

marcoabreu commented on a change in pull request #9862: Fix a race condition in 
converting data layouts in MKLDNN.
URL: https://github.com/apache/incubator-mxnet/pull/9862#discussion_r170113380
 
 

 ##
 File path: tests/python/gpu/test_gluon_model_zoo_gpu.py
 ##
 @@ -105,6 +107,7 @@ def test_training():
 # TODO(zhengda) mobilenet can't pass this test even without MKLDNN.
 all_models = ['resnet18_v1', 'densenet121']
 
+mx.random.seed(1)
 
 Review comment:
   please don't use mx.random.seed due to the introduction of the @with_seed() 
decorator


This is an automated message from the Apache Git Service.
To respond to the message, please log on GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


With regards,
Apache Git Services

[GitHub] zheng-da commented on issue #9843: Flaky test_random.test_exponential_generator @ Python2: MKLDNN-CPU

[GitBox]

zheng-da commented on issue #9843: Flaky test_random.test_exponential_generator 
@ Python2: MKLDNN-CPU
URL: 
https://github.com/apache/incubator-mxnet/issues/9843#issuecomment-367840947
 
 
   It seems this problem is related to 
https://github.com/apache/incubator-mxnet/issues/9856
   


This is an automated message from the Apache Git Service.
To respond to the message, please log on GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


With regards,
Apache Git Services

[GitHub] zheng-da commented on issue #9856: Flaky test_random.test_gamma_generator

[GitBox]

zheng-da commented on issue #9856: Flaky test_random.test_gamma_generator
URL: 
https://github.com/apache/incubator-mxnet/issues/9856#issuecomment-367839764
 
 
   Is this problem related to MKLDNN?


This is an automated message from the Apache Git Service.
To respond to the message, please log on GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


With regards,
Apache Git Services

[GitHub] cjolivier01 commented on issue #9858: softsign activation function

[GitBox]

cjolivier01 commented on issue #9858: softsign activation function
URL: https://github.com/apache/incubator-mxnet/pull/9858#issuecomment-367833441
 
 
   "whenever it's convenient"?  none of this seems terribly "convenient".  Not 
sure where that framing came from.  Customer obsessions is rarely about 
convenience.
   Anyway, contrib sounds fine, to me.


This is an automated message from the Apache Git Service.
To respond to the message, please log on GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


With regards,
Apache Git Services

[GitHub] marcoabreu commented on issue #9809: fix optimizer bug in CPP-Package

[GitBox]

marcoabreu commented on issue #9809: fix optimizer bug in CPP-Package
URL: https://github.com/apache/incubator-mxnet/pull/9809#issuecomment-367830881
 
 
   @larroy 


This is an automated message from the Apache Git Service.
To respond to the message, please log on GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


With regards,
Apache Git Services

[GitHub] zheng-da opened a new pull request #9862: Fix a race condition in converting data layouts.

[GitBox]

zheng-da opened a new pull request #9862: Fix a race condition in converting 
data layouts.
URL: https://github.com/apache/incubator-mxnet/pull/9862
 
 
   ## Description ##
   There is a race condition in data layout conversion in the MKLDNN 
implementation. 
   Currently, when NDArray::data() is called, it converts data format (from the 
MKLDNN format to the default format) inside an NDArray. In the threaded 
execution engine, while an NDArray is being converted in a thread, the array 
can also be read by another thread. In this case, the other thread can read 
wrong data. A similar race condition may also exist in the MKLML integration, 
which sometimes generates the garbage results.
   
   Please see the details of the error in 
https://github.com/apache/incubator-mxnet/issues/9820
   
   ## Checklist ##
   ### Essentials ###
   - [ ] Passed code style checking (`make lint`)
   - [ ] Changes are complete (i.e. I finished coding on this PR)
   - [ ] All changes have test coverage:
   - Unit tests are added for small changes to verify correctness (e.g. adding 
a new operator)
   - Nightly tests are added for complicated/long-running ones (e.g. changing 
distributed kvstore)
   - Build tests will be added for build configuration changes (e.g. adding a 
new build option with NCCL)
   - [ ] Code is well-documented: 
   - For user-facing API changes, API doc string has been updated. 
   - For new C++ functions in header files, their functionalities and arguments 
are documented. 
   - For new examples, README.md is added to explain the what the example does, 
the source of the dataset, expected performance on test set and reference to 
the original paper if applicable
   - [ ] To the my best knowledge, examples are either not affected by this 
change, or have been fixed to be compatible with this change
   


This is an automated message from the Apache Git Service.
To respond to the message, please log on GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


With regards,
Apache Git Services

[GitHub] sifmelcara commented on issue #9809: fix optimizer bug in CPP-Package

[GitBox]

sifmelcara commented on issue #9809: fix optimizer bug in CPP-Package
URL: https://github.com/apache/incubator-mxnet/pull/9809#issuecomment-367824559
 
 
   Hmm, in fact I think adding back `static int __make_ ## OptimizerType ## _ 
## Name ## __ =` is a little bit better because local static variable's 
initialization is guaranteed to be thread safe.


This is an automated message from the Apache Git Service.
To respond to the message, please log on GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


With regards,
Apache Git Services

[GitHub] sifmelcara commented on issue #9809: fix optimizer bug in CPP-Package

[GitBox]

sifmelcara commented on issue #9809: fix optimizer bug in CPP-Package
URL: https://github.com/apache/incubator-mxnet/pull/9809#issuecomment-367822502
 
 
   Hello! Just got a ping from @chsin for this PR. I'm the author of the ODR 
fix.
   After some git blame, I found this bug was accidentally introduced by 
https://github.com/apache/incubator-mxnet/commit/1c22bc71052f54eb21a6079eebdc7383dafbb4ee
 which tries to fix a compiler warning of return value not used... But we need 
`static int __make_ ## OptimizerType ## _ ## Name ## __ =` for a reason: it is 
used to make sure optimizers are only registered once.
   So either add back `static int __make_ ## OptimizerType ## _ ## Name ## __ 
=` or merge this PR should be able to fix #9800 .
   
   However, as @szha mentioned, I wonder if there is a better way to register 
the optimizers? I guess we can use global inline variable but that would 
require C++17


This is an automated message from the Apache Git Service.
To respond to the message, please log on GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


With regards,
Apache Git Services

[GitHub] sifmelcara commented on issue #9809: fix optimizer bug in CPP-Package

[GitBox]

sifmelcara commented on issue #9809: fix optimizer bug in CPP-Package
URL: https://github.com/apache/incubator-mxnet/pull/9809#issuecomment-367822502
 
 
   Hello! Just got a ping from @chsin for this PR. I'm the author of the ODR 
fix.
   After some git blame, I found this bug was accidentally introduced by 
https://github.com/apache/incubator-mxnet/commit/1c22bc71052f54eb21a6079eebdc7383dafbb4ee
 which tries to fix a compiler warning of return value not used... But we need 
`static int __make_ ## OptimizerType ## _ ## Name ## __ =` for a reason, it is 
used to make sure optimizers are only registered once.
   So either add back `static int __make_ ## OptimizerType ## _ ## Name ## __ 
=` or merge this PR should be able to fix #9800 .
   
   However, as @szha mentioned, I wonder if there is a better way to register 
the optimizers? I guess we can use global inline variable but that would 
require C++17


This is an automated message from the Apache Git Service.
To respond to the message, please log on GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


With regards,
Apache Git Services

[GitHub] sifmelcara commented on issue #9809: fix optimizer bug in CPP-Package

[GitBox]

sifmelcara commented on issue #9809: fix optimizer bug in CPP-Package
URL: https://github.com/apache/incubator-mxnet/pull/9809#issuecomment-367822502
 
 
   Hello! Just got a ping from @chsin for this PR. I'm the author of the ODR 
fix.
   After some git blame, I found this bug was accidentally introduced by 
https://github.com/apache/incubator-mxnet/commit/1c22bc71052f54eb21a6079eebdc7383dafbb4ee
 which tries to fix a compiler warning of return value not used... We need 
`static int __make_ ## OptimizerType ## _ ## Name ## __ =` to make sure 
optimizers are only registered once.
   So either add back `static int __make_ ## OptimizerType ## _ ## Name ## __ 
=` or merge this PR should be able to fix #9800 .
   
   However, as @szha mentioned, I wonder if there is a better way to register 
the optimizers? I guess we can use global inline variable but that would 
require C++17


This is an automated message from the Apache Git Service.
To respond to the message, please log on GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


With regards,
Apache Git Services

[GitHub] szha commented on issue #9858: softsign activation function

[GitBox]

szha commented on issue #9858: softsign activation function
URL: https://github.com/apache/incubator-mxnet/pull/9858#issuecomment-367820170
 
 
   The rules are standards we set for ourselves, and we probably don't want 
people to think this is a community that's OK with bending the rules whenever 
it's more convenient.
   
   Under these constraints, I'd recommend adding the API to v0.11 only as a 
contrib operator. If you want it to appear as the official api, then it's 
necessary to stick to what Eric suggested and add it to all successive versions.


This is an automated message from the Apache Git Service.
To respond to the message, please log on GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


With regards,
Apache Git Services

[GitHub] szha commented on issue #9858: softsign activation function

[GitBox]

szha commented on issue #9858: softsign activation function
URL: https://github.com/apache/incubator-mxnet/pull/9858#issuecomment-367820170
 
 
   The rules are standards we set for ourselves, and we probably don't want 
people to think this is a community that's OK to bend the rules whenever it's 
more convenient.
   
   Under these constraints, I'd recommend adding the API to v0.11 only as a 
contrib operator. If you want it to appear as the official api, then it's 
necessary to stick to what Eric suggested and add it to all successive versions.


This is an automated message from the Apache Git Service.
To respond to the message, please log on GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


With regards,
Apache Git Services

[GitHub] cjolivier01 commented on issue #9858: softsign activation function

[GitBox]

cjolivier01 commented on issue #9858: softsign activation function
URL: https://github.com/apache/incubator-mxnet/pull/9858#issuecomment-367818944
 
 
   probably this discussion should occur on dev and see what the mentors think
   
   On Thu, Feb 22, 2018 at 12:42 PM, Marco de Abreu 
   wrote:
   
   > Of course, but we can't guarantee correctness due to lacking CI and we'd
   > have to make a formal release according to Apache in order to be able to
   > supply new binaries.
   >
   > ?
   > You are receiving this because you were mentioned.
   > Reply to this email directly, view it on GitHub
   > 
,
   > or mute the thread
   > 

   > .
   >
   


This is an automated message from the Apache Git Service.
To respond to the message, please log on GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


With regards,
Apache Git Services

[GitHub] marcoabreu commented on issue #9858: softsign activation function

[GitBox]

marcoabreu commented on issue #9858: softsign activation function
URL: https://github.com/apache/incubator-mxnet/pull/9858#issuecomment-367815373
 
 
   Of course, but we can't guarantee correctness due to lacking CI and we'd 
have to make a formal release according to Apache in order to be able to supply 
new binaries. 


This is an automated message from the Apache Git Service.
To respond to the message, please log on GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


With regards,
Apache Git Services