[jira] [Assigned] (MXNET-216) The method reset() of DataIter does not work

[ https://issues.apache.org/jira/browse/MXNET-216?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Rahul Huilgol reassigned MXNET-216: --- Assignee: (was: Anirudh Subramanian) > The method reset() of DataIter does not work >

[jira] [Assigned] (MXNET-116) Object Detector using classifier and ImageClassifier functions

[ https://issues.apache.org/jira/browse/MXNET-116?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Qing Lan reassigned MXNET-116: -- Assignee: Qing Lan > Object Detector using classifier and ImageClassifier functions > --

[jira] [Assigned] (MXNET-256) Scala GPU tests CI setup

[ https://issues.apache.org/jira/browse/MXNET-256?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Qing Lan reassigned MXNET-256: -- Assignee: Qing Lan > Scala GPU tests CI setup > > > Key: MXNET-

[jira] [Updated] (MXNET-93) Sparse support for Custom Ops

[ https://issues.apache.org/jira/browse/MXNET-93?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Anirudh Subramanian updated MXNET-93: - Status: In Review (was: In Progress) > Sparse support for Custom Ops >

[jira] [Updated] (MXNET-158) Training with the same parameters and seed gets significantly different results

[ https://issues.apache.org/jira/browse/MXNET-158?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Chris Olivier updated MXNET-158: Status: In Progress (was: To Do) > Training with the same parameters and seed gets significantly dif

[jira] [Commented] (MXNET-187) Support pre-kepler GPUs without __shfl_down instruction

[ https://issues.apache.org/jira/browse/MXNET-187?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16423211#comment-16423211 ] Chris Olivier commented on MXNET-187: - PR: https://github.com/apache/incubator-mxnet/pu

[jira] [Updated] (MXNET-23) Profiling server process in distributed training

[ https://issues.apache.org/jira/browse/MXNET-23?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Rahul Huilgol updated MXNET-23: --- Status: In Review (was: In Progress) > Profiling server process in distributed training > -

[jira] [Updated] (MXNET-32) Improve documentation of environment variables

[ https://issues.apache.org/jira/browse/MXNET-32?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Rahul Huilgol updated MXNET-32: --- Resolution: Fixed Status: Done (was: To Do) > Improve documentation of environment variables >

[jira] [Updated] (MXNET-36) Update ndarray binary ops to use kernel launch interface

[ https://issues.apache.org/jira/browse/MXNET-36?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Rahul Huilgol updated MXNET-36: --- Resolution: Fixed Status: Done (was: To Do) > Update ndarray binary ops to use kernel launch in

[jira] [Updated] (MXNET-37) Tutorial for distributed training

[ https://issues.apache.org/jira/browse/MXNET-37?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Rahul Huilgol updated MXNET-37: --- Status: In Progress (was: To Do) > Tutorial for distributed training >

[jira] [Updated] (MXNET-37) Tutorial for distributed training

[ https://issues.apache.org/jira/browse/MXNET-37?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Rahul Huilgol updated MXNET-37: --- Status: In Review (was: In Progress) > Tutorial for distributed training >

[jira] [Updated] (MXNET-120) FP16 support for distributed training

[ https://issues.apache.org/jira/browse/MXNET-120?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Rahul Huilgol updated MXNET-120: Status: In Review (was: In Progress) > FP16 support for distributed training > -

[jira] [Updated] (MXNET-156) IndexError in labels when size of training dataset is not multiple of batch size

[ https://issues.apache.org/jira/browse/MXNET-156?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Rahul Huilgol updated MXNET-156: Priority: Minor (was: Critical) > IndexError in labels when size of training dataset is not multiple

[jira] [Created] (MXNET-264) improve the performance of MKLDNN for small batch sizes.

Da Zheng created MXNET-264: -- Summary: improve the performance of MKLDNN for small batch sizes. Key: MXNET-264 URL: https://issues.apache.org/jira/browse/MXNET-264 Project: Apache MXNet Issue Type: I

[jira] [Updated] (MXNET-187) Support pre-kepler GPUs without __shfl_down instruction

[ https://issues.apache.org/jira/browse/MXNET-187?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Chris Olivier updated MXNET-187: Status: In Progress (was: To Do) > Support pre-kepler GPUs without __shfl_down instruction > ---

[jira] [Commented] (MXNET-194) Fatal error on dot operation between dense and sparse vectors

[ https://issues.apache.org/jira/browse/MXNET-194?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16423050#comment-16423050 ] Hao Jin commented on MXNET-194: --- To address this issue and the related issue(#8817), we will

[jira] [Assigned] (MXNET-156) IndexError in labels when size of training dataset is not multiple of batch size

[ https://issues.apache.org/jira/browse/MXNET-156?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Hao Jin reassigned MXNET-156: - Assignee: Hao Jin > IndexError in labels when size of training dataset is not multiple of batch > size >

[jira] [Assigned] (MXNET-170) Call symbol/nd.smooth_l1 without scalar causing uncaught exception

[ https://issues.apache.org/jira/browse/MXNET-170?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Rahul Huilgol reassigned MXNET-170: --- Assignee: Rahul Huilgol > Call symbol/nd.smooth_l1 without scalar causing uncaught exception >

[jira] [Assigned] (MXNET-161) gluon hang-up unexpectedly

[ https://issues.apache.org/jira/browse/MXNET-161?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Rahul Huilgol reassigned MXNET-161: --- Assignee: Rahul Huilgol > gluon hang-up unexpectedly > -- > >

[jira] [Assigned] (MXNET-156) IndexError in labels when size of training dataset is not multiple of batch size

[ https://issues.apache.org/jira/browse/MXNET-156?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Rahul Huilgol reassigned MXNET-156: --- Assignee: Hao Jin (was: Rahul Huilgol) > IndexError in labels when size of training dataset i

[jira] [Assigned] (MXNET-188) problem when set fix_gamma=True in batchnorm

[ https://issues.apache.org/jira/browse/MXNET-188?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Rahul Huilgol reassigned MXNET-188: --- Assignee: Rahul Huilgol > problem when set fix_gamma=True in batchnorm > -

[jira] [Assigned] (MXNET-156) IndexError in labels when size of training dataset is not multiple of batch size

[ https://issues.apache.org/jira/browse/MXNET-156?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Rahul Huilgol reassigned MXNET-156: --- Assignee: (was: Hao Jin) > IndexError in labels when size of training dataset is not multi

[jira] [Assigned] (MXNET-156) IndexError in labels when size of training dataset is not multiple of batch size

[ https://issues.apache.org/jira/browse/MXNET-156?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Rahul Huilgol reassigned MXNET-156: --- Assignee: Rahul Huilgol > IndexError in labels when size of training dataset is not multiple o

[jira] [Assigned] (MXNET-159) Segfault on ndarray with negative dimension i.e. mxnet.nx.zeros((-1,))

[ https://issues.apache.org/jira/browse/MXNET-159?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Rahul Huilgol reassigned MXNET-159: --- Assignee: Rahul Huilgol > Segfault on ndarray with negative dimension i.e. mxnet.nx.zeros((-1,

[jira] [Assigned] (MXNET-155) ndarray indexing issues

[ https://issues.apache.org/jira/browse/MXNET-155?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Rahul Huilgol reassigned MXNET-155: --- Assignee: Jun Wu > ndarray indexing issues > --- > > Key:

[jira] [Assigned] (MXNET-182) precision bug in batchnorm.

[ https://issues.apache.org/jira/browse/MXNET-182?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Rahul Huilgol reassigned MXNET-182: --- Assignee: Jun Wu > precision bug in batchnorm. > --- > >

[jira] [Assigned] (MXNET-225) Dangling outputs and dtype != float32: Gradient computation fails

[ https://issues.apache.org/jira/browse/MXNET-225?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Rahul Huilgol reassigned MXNET-225: --- Assignee: Jun Wu > Dangling outputs and dtype != float32: Gradient computation fails > ---

[jira] [Assigned] (MXNET-162) Backward shape inconsistent with custom HybridBlock and gluon.loss

[ https://issues.apache.org/jira/browse/MXNET-162?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Rahul Huilgol reassigned MXNET-162: --- Assignee: Jun Wu > Backward shape inconsistent with custom HybridBlock and gluon.loss > --

[jira] [Assigned] (MXNET-157) Gluon Conv1DTranspose causes "kernel died"

[ https://issues.apache.org/jira/browse/MXNET-157?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Rahul Huilgol reassigned MXNET-157: --- Assignee: Hao Jin > Gluon Conv1DTranspose causes "kernel died" > -

[jira] [Assigned] (MXNET-232) mx random seed doesn't work for random_uniform/random_normal on gpu

[ https://issues.apache.org/jira/browse/MXNET-232?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Rahul Huilgol reassigned MXNET-232: --- Assignee: Hao Jin > mx random seed doesn't work for random_uniform/random_normal on gpu >

[jira] [Assigned] (MXNET-165) mxnet-mkl (v0.12.0) crash when using (conda-installed) numpy with MKL

[ https://issues.apache.org/jira/browse/MXNET-165?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Rahul Huilgol reassigned MXNET-165: --- Assignee: Hao Jin > mxnet-mkl (v0.12.0) crash when using (conda-installed) numpy with MKL > --

[jira] [Assigned] (MXNET-176) Reduce operators do not support axis=None

[ https://issues.apache.org/jira/browse/MXNET-176?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Rahul Huilgol reassigned MXNET-176: --- Assignee: Haibin Lin > Reduce operators do not support axis=None > ---

[jira] [Assigned] (MXNET-160) sample_multinomial crashes on vector input

[ https://issues.apache.org/jira/browse/MXNET-160?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Rahul Huilgol reassigned MXNET-160: --- Assignee: Hao Jin > sample_multinomial crashes on vector input > -

[jira] [Assigned] (MXNET-194) Fatal error on dot operation between dense and sparse vectors

[ https://issues.apache.org/jira/browse/MXNET-194?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Rahul Huilgol reassigned MXNET-194: --- Assignee: Hao Jin > Fatal error on dot operation between dense and sparse vectors > --

[jira] [Assigned] (MXNET-154) Inconsistent weight decay logics in multiple optimizers

[ https://issues.apache.org/jira/browse/MXNET-154?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Rahul Huilgol reassigned MXNET-154: --- Assignee: Haibin Lin > Inconsistent weight decay logics in multiple optimizers > -

[jira] [Assigned] (MXNET-205) simple_bind elemwise_add with group2ctx fails

[ https://issues.apache.org/jira/browse/MXNET-205?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Rahul Huilgol reassigned MXNET-205: --- Assignee: Haibin Lin > simple_bind elemwise_add with group2ctx fails > ---

[jira] [Assigned] (MXNET-158) Training with the same parameters and seed gets significantly different results

[ https://issues.apache.org/jira/browse/MXNET-158?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Rahul Huilgol reassigned MXNET-158: --- Assignee: Chris Olivier > Training with the same parameters and seed gets significantly differ

[jira] [Assigned] (MXNET-172) (info.type) != (kNotInitialized)

[ https://issues.apache.org/jira/browse/MXNET-172?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Rahul Huilgol reassigned MXNET-172: --- Assignee: Haibin Lin > (info.type) != (kNotInitialized) > > >

[jira] [Assigned] (MXNET-193) The custom operator not supported for group context?

[ https://issues.apache.org/jira/browse/MXNET-193?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Rahul Huilgol reassigned MXNET-193: --- Assignee: Haibin Lin > The custom operator not supported for group context? >

[jira] [Assigned] (MXNET-214) BatchNorm Evaluation Mode Backward Fails with cudnn Enabled

[ https://issues.apache.org/jira/browse/MXNET-214?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Rahul Huilgol reassigned MXNET-214: --- Assignee: Chris Olivier > BatchNorm Evaluation Mode Backward Fails with cudnn Enabled > --

[jira] [Assigned] (MXNET-187) Support pre-kepler GPUs without __shfl_down instruction

[ https://issues.apache.org/jira/browse/MXNET-187?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Rahul Huilgol reassigned MXNET-187: --- Assignee: Chris Olivier > Support pre-kepler GPUs without __shfl_down instruction > --

[jira] [Assigned] (MXNET-207) incompatibility between backward mirror and batch normalization

[ https://issues.apache.org/jira/browse/MXNET-207?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Rahul Huilgol reassigned MXNET-207: --- Assignee: Chris Olivier > incompatibility between backward mirror and batch normalization > --

[jira] [Assigned] (MXNET-181) Asynchronous Issue on CustomOP and mxnet.image.ImageDetIter

[ https://issues.apache.org/jira/browse/MXNET-181?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Rahul Huilgol reassigned MXNET-181: --- Assignee: Anirudh Subramanian > Asynchronous Issue on CustomOP and mxnet.image.ImageDetIter >

[jira] [Assigned] (MXNET-153) No training happening when CSVIter() is used.

[ https://issues.apache.org/jira/browse/MXNET-153?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Rahul Huilgol reassigned MXNET-153: --- Assignee: Anirudh Subramanian > No training happening when CSVIter() is used. > --

[jira] [Assigned] (MXNET-199) Cast storage to same stype gives an error for csr

[ https://issues.apache.org/jira/browse/MXNET-199?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Rahul Huilgol reassigned MXNET-199: --- Assignee: Anirudh Subramanian > Cast storage to same stype gives an error for csr > --

[jira] [Assigned] (MXNET-201) different behaviour of customop in latest MXNet

[ https://issues.apache.org/jira/browse/MXNET-201?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Rahul Huilgol reassigned MXNET-201: --- Assignee: Anirudh Subramanian > different behaviour of customop in latest MXNet >

[jira] [Assigned] (MXNET-224) Cannot find custom operator when visualizing network.

[ https://issues.apache.org/jira/browse/MXNET-224?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Rahul Huilgol reassigned MXNET-224: --- Assignee: Anirudh Subramanian > Cannot find custom operator when visualizing network. > --

[jira] [Assigned] (MXNET-216) The method reset() of DataIter does not work

[ https://issues.apache.org/jira/browse/MXNET-216?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Rahul Huilgol reassigned MXNET-216: --- Assignee: Anirudh Subramanian > The method reset() of DataIter does not work > ---

[jira] [Assigned] (MXNET-223) src/operator/./bilinear_sampler-inl.h:105: Have not implemented the data req combinations! gdata_req=0 ggrid_req=1

[ https://issues.apache.org/jira/browse/MXNET-223?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Rahul Huilgol reassigned MXNET-223: --- Assignee: Chris Olivier > src/operator/./bilinear_sampler-inl.h:105: Have not implemented the

[jira] [Updated] (MXNET-104) PythonLibs missing from build slave

[ https://issues.apache.org/jira/browse/MXNET-104?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Chris Olivier updated MXNET-104: Resolution: Fixed Status: Done (was: To Do) > PythonLibs missing from build slave >

[jira] [Updated] (MXNET-134) Don't do imperative bulk when profiling aggregate stats

[ https://issues.apache.org/jira/browse/MXNET-134?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Chris Olivier updated MXNET-134: Status: Done (was: In Progress) > Don't do imperative bulk when profiling aggregate stats >

[jira] [Updated] (MXNET-88) Test and time GetContext() call rewritten in cython

[ https://issues.apache.org/jira/browse/MXNET-88?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Chris Olivier updated MXNET-88: --- Status: To Do (was: In Progress) > Test and time GetContext() call rewritten in cython > --

[jira] [Created] (MXNET-263) dot(dense, csr.T) = dense on GPU

Hao Jin created MXNET-263: - Summary: dot(dense, csr.T) = dense on GPU Key: MXNET-263 URL: https://issues.apache.org/jira/browse/MXNET-263 Project: Apache MXNet Issue Type: New Feature Rep

[jira] [Assigned] (MXNET-263) dot(dense, csr.T) = dense on GPU

[ https://issues.apache.org/jira/browse/MXNET-263?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Hao Jin reassigned MXNET-263: - Assignee: Hao Jin > dot(dense, csr.T) = dense on GPU > > >

[jira] [Updated] (MXNET-230) MKL Bug

[ https://issues.apache.org/jira/browse/MXNET-230?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Rahul Huilgol updated MXNET-230: Status: In Progress (was: To Do) > MKL Bug > --- > > Key: MXNET-230 >

[jira] [Updated] (MXNET-230) MKL Bug

[ https://issues.apache.org/jira/browse/MXNET-230?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Rahul Huilgol updated MXNET-230: Status: Done (was: In Progress) > MKL Bug > --- > > Key: MXNET-230 >

[jira] [Commented] (MXNET-230) MKL Bug

[ https://issues.apache.org/jira/browse/MXNET-230?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16422804#comment-16422804 ] Rahul Huilgol commented on MXNET-230: - Looks like MKLDNN doesn't have this issue. Closi

[jira] [Assigned] (MXNET-231) Bug in backprop for the transpose of a transpose

[ https://issues.apache.org/jira/browse/MXNET-231?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Rahul Huilgol reassigned MXNET-231: --- Assignee: Jun Wu > Bug in backprop for the transpose of a transpose >

[jira] [Updated] (MXNET-72) Improve sparse optimizer on GPU

[ https://issues.apache.org/jira/browse/MXNET-72?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Haibin Lin updated MXNET-72: Status: In Review (was: In Progress) > Improve sparse optimizer on GPU > --- > >

[jira] [Updated] (MXNET-144) Fix test_sparse_quadratic_function test

[ https://issues.apache.org/jira/browse/MXNET-144?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Haibin Lin updated MXNET-144: - Status: Done (was: In Progress) > Fix test_sparse_quadratic_function test > --

[jira] [Updated] (MXNET-241) Add module api for distributed training with rowsparse weight

[ https://issues.apache.org/jira/browse/MXNET-241?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Haibin Lin updated MXNET-241: - Sprint: Apache Sprint 2 > Add module api for distributed training with rowsparse weight > -

[jira] [Updated] (MXNET-144) Fix test_sparse_quadratic_function test

[ https://issues.apache.org/jira/browse/MXNET-144?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Haibin Lin updated MXNET-144: - Sprint: Apache Sprint 2 > Fix test_sparse_quadratic_function test > ---

[jira] [Updated] (MXNET-72) Improve sparse optimizer on GPU

[ https://issues.apache.org/jira/browse/MXNET-72?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Haibin Lin updated MXNET-72: Sprint: Apache Sprint 2 > Improve sparse optimizer on GPU > --- > >

[jira] [Updated] (MXNET-241) Add module api for distributed training with rowsparse weight

[ https://issues.apache.org/jira/browse/MXNET-241?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Haibin Lin updated MXNET-241: - Assignee: Haibin Lin Status: In Progress (was: To Do) > Add module api for distributed training with

[jira] [Created] (MXNET-262) Implement mx.random.seed_context to seed random number generators of a specific device context

Deokjae Lee created MXNET-262: - Summary: Implement mx.random.seed_context to seed random number generators of a specific device context Key: MXNET-262 URL: https://issues.apache.org/jira/browse/MXNET-262

[jira] [Updated] (MXNET-87) Add JIRA-related guidelines to docs

[ https://issues.apache.org/jira/browse/MXNET-87?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Chris Olivier updated MXNET-87: --- Status: Done (was: In Review) > Add JIRA-related guidelines to docs > -

[jira] [Updated] (MXNET-87) Add JIRA-related guidelines to docs

[ https://issues.apache.org/jira/browse/MXNET-87?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Chris Olivier updated MXNET-87: --- Status: In Review (was: In Progress) > Add JIRA-related guidelines to docs > --

[jira] [Created] (MXNET-261) Update MKLDNN & Add CPP Test

Xinyu Chen created MXNET-261: Summary: Update MKLDNN & Add CPP Test Key: MXNET-261 URL: https://issues.apache.org/jira/browse/MXNET-261 Project: Apache MXNet Issue Type: Improvement R