dongzhuoyao opened a new pull request #9641: sgld bugfix
URL: https://github.com/apache/incubator-mxnet/pull/9641
## Description ##
fix issues in #https://github.com/apache/incubator-mxnet/issues/9605 about
SGLD
## Checklist ##
### Essentials ###
- [Y ] Passed code style
feiyuvl commented on issue #9611: program can't finished normally in dist_sync
mode
URL:
https://github.com/apache/incubator-mxnet/issues/9611#issuecomment-361849189
For distributed training, it's better to set the total iteration of each
node to the same number. The previous model provid
feiyuvl commented on issue #9629: do not save gpu memory during fp16 training
URL:
https://github.com/apache/incubator-mxnet/issues/9629#issuecomment-361848727
Maybe related to your network structure, we tested fp16 on cnn/lstm, it can
reduce gpu memory cost
--
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 a0c8224 Update CONTRIBUTORS.md (#963
szha opened a new issue #9640: mx.metric lacks necessary tests for correctness
URL: https://github.com/apache/incubator-mxnet/issues/9640
Currently no actual test is performed for metrics other than nll loss.
This is an automa
szha closed pull request #9638: Update CONTRIBUTORS.md
URL: https://github.com/apache/incubator-mxnet/pull/9638
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 (fr
szha commented on a change in pull request #9636: avoid per-batch blocking in
metric
URL: https://github.com/apache/incubator-mxnet/pull/9636#discussion_r164970146
##
File path: python/mxnet/metric.py
##
@@ -399,8 +401,10 @@ def update(self, labels, preds):
if
This is an automated email from the ASF dual-hosted git repository.
jxie 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 c8241cb Update test_kvstore.py (#9639)
c
gzh1991 commented on issue #9537: float16 have no performance improvemen in
training cifar10
URL:
https://github.com/apache/incubator-mxnet/issues/9537#issuecomment-361837538
Thanks a lot for your reply and explanation! I have tested on Titan X as you
said, and fp16 do have a better perfo
marcoabreu commented on issue #9639: Add back kvstore unit test
URL: https://github.com/apache/incubator-mxnet/pull/9639#issuecomment-361833690
Is the fact that the test runs on a GPU the problem or just the test itself?
This
eric-haibin-lin opened a new pull request #9639: Add back kvstore unit test
URL: https://github.com/apache/incubator-mxnet/pull/9639
`test_row_sparse_pull` is working correctly on CPU. It was mistakenly
disabled by a previous PR. The non-working one is in `gpu/test_kvstore.py`,
which is di
szha commented on issue #9633: Gluon image-classification example improvement
URL: https://github.com/apache/incubator-mxnet/pull/9633#issuecomment-361831216
maybe move it to mx.contrib.io first?
This is an automated message f
315386775 commented on issue #9629: do not save gpu memory during fp16 training
URL:
https://github.com/apache/incubator-mxnet/issues/9629#issuecomment-361829519
how to training with fp16 in detection. like faster-rcnn @tornadomeet
-
dongzhuoyao opened a new pull request #9638: Update CONTRIBUTORS.md
URL: https://github.com/apache/incubator-mxnet/pull/9638
## Description ##
(Brief description on what this PR is about)
## Checklist ##
### Essentials ###
- [Y] Passed code style checking (`make lint`)
- [
reminisce commented on issue #9637: input with name 'data' is not found in
symbol.list_arguments()
URL:
https://github.com/apache/incubator-mxnet/issues/9637#issuecomment-361823101
What's the output of symbol.list_arguments()? You need to use the data name
in the outputs of list_arguments
cjolivier01 closed pull request #9389: [SUBMODULE] Upgrade ps-lite to the
latest version
URL: https://github.com/apache/incubator-mxnet/pull/9389
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 an automated email from the ASF dual-hosted git repository.
cjolivier01 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 be39bcf [SUBMODULE] Upgrade ps-li
zhreshold commented on issue #9633: Gluon image-classification example
improvement
URL: https://github.com/apache/incubator-mxnet/pull/9633#issuecomment-361819957
@szha Too many lines for mx.io module, suggestions?
This is an
VictoriaYyz opened a new issue #9637: input with name 'data' is not found in
symbol.list_arguments()
URL: https://github.com/apache/incubator-mxnet/issues/9637
I try to construct a cnn model with MxNet Symbol API (1.0), but when I
create a module:
mod =
mx.mod.Module(symbol=out,label_n
CodingCat commented on issue #9389: [SUBMODULE] Upgrade ps-lite to the latest
version
URL: https://github.com/apache/incubator-mxnet/pull/9389#issuecomment-361818280
@cjolivier01 @eric-haibin-lin @yzhliu since we have cut out 1.1 branch, is
it ready to merge this to master?
--
piiswrong closed pull request #9481: add where op with sparse condition
URL: https://github.com/apache/incubator-mxnet/pull/9481
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
This is an automated email from the ASF dual-hosted git repository.
jxie 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 8e9798c add where op with sparse conditi
TaoLv commented on issue #9545: Profiling discussion
URL:
https://github.com/apache/incubator-mxnet/issues/9545#issuecomment-361808896
> " if there is no operator parallelism, maybe we should give as many
threads as possible to the running operator, especially for mkldnn operator"
>
TaoLv commented on issue #9545: Profiling discussion
URL:
https://github.com/apache/incubator-mxnet/issues/9545#issuecomment-361807553
> "After set USE_OPERATOR_TUNING=0 and rebuilt mxnet, the first 56 threads
were not there anymore. Do you think this behaviour is as expectaton?"
> WHat
TaoLv commented on issue #9545: Profiling discussion
URL:
https://github.com/apache/incubator-mxnet/issues/9545#issuecomment-361807553
> "After set USE_OPERATOR_TUNING=0 and rebuilt mxnet, the first 56 threads
were not there anymore. Do you think this behaviour is as expectaton?"
> WHat
TaoLv commented on issue #9152: tutorial for distributed training
URL: https://github.com/apache/incubator-mxnet/pull/9152#issuecomment-361806976
Any content about how to profile the time of operator execution and
communication and how well they are overlapped?
yanhn commented on issue #9612: CUDNN_STATUS_SUCCESS (4 vs. 0) cuDNN:
CUDNN_STATUS_INTERNAL_ERROR on jetson TX2
URL:
https://github.com/apache/incubator-mxnet/issues/9612#issuecomment-361805409
hi @larroy?thank you for replying. I tried ssd demo `python demo.py --gpu 0`
under path `XX/inc
ray3572 closed issue #8038: why I use MXPredCreate(), it shows "
mxnet_predict-all.cc:23570: Check failed: op != nullptr Operator
RegressionOutput is not registered"?
URL: https://github.com/apache/incubator-mxnet/issues/8038
-
szha closed issue #9332: "Matrix Factorization" R test failing due to
unavailable HTTP server.
URL: https://github.com/apache/incubator-mxnet/issues/9332
This is an automated message from the Apache Git Service.
To respond t
szha commented on a change in pull request #9633: Gluon image-classification
example improvement
URL: https://github.com/apache/incubator-mxnet/pull/9633#discussion_r164931533
##
File path: example/gluon/image_classification.py
##
@@ -17,36 +17,49 @@
from __future__ imp
szha commented on a change in pull request #9633: Gluon image-classification
example improvement
URL: https://github.com/apache/incubator-mxnet/pull/9633#discussion_r164932130
##
File path: example/gluon/image_classification.py
##
@@ -62,63 +75,85 @@
h
szha commented on a change in pull request #9633: Gluon image-classification
example improvement
URL: https://github.com/apache/incubator-mxnet/pull/9633#discussion_r164932515
##
File path: example/gluon/image_classification.py
##
@@ -160,23 +207,34 @@ def train(epochs, ct
szha commented on a change in pull request #9633: Gluon image-classification
example improvement
URL: https://github.com/apache/incubator-mxnet/pull/9633#discussion_r164932765
##
File path: example/gluon/image_classification.py
##
@@ -160,23 +207,34 @@ def train(epochs, ct
szha commented on a change in pull request #9633: Gluon image-classification
example improvement
URL: https://github.com/apache/incubator-mxnet/pull/9633#discussion_r164932256
##
File path: example/gluon/image_classification.py
##
@@ -62,63 +75,85 @@
h
szha commented on a change in pull request #9633: Gluon image-classification
example improvement
URL: https://github.com/apache/incubator-mxnet/pull/9633#discussion_r164932389
##
File path: example/gluon/image_classification.py
##
@@ -129,27 +164,39 @@ def test(ctx, val_da
zheng-da commented on a change in pull request #8302: Refactor operators &
MKLDNN
URL: https://github.com/apache/incubator-mxnet/pull/8302#discussion_r164930373
##
File path: tests/python/gpu/test_gluon_model_zoo.py
##
@@ -0,0 +1,163 @@
+# Licensed to the Apache Software F
szha closed pull request #9630: Fix the version on website mainpage
URL: https://github.com/apache/incubator-mxnet/pull/9630
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 pul
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 87ba915 Fix the version on website m
cjolivier01 commented on issue #9615: [CMake] Fix OSX double compilation and
add static library
URL: https://github.com/apache/incubator-mxnet/pull/9615#issuecomment-361788898
please note that "incorrect" is not a valid description. Please replace
"incorrect" with an actual description of
cjolivier01 commented on issue #9615: [CMake] Fix OSX double compilation and
add static library
URL: https://github.com/apache/incubator-mxnet/pull/9615#issuecomment-361788898
please note that "incorrect" is not a valid description. Please replace
"incorrect" with an actual description of
eric-haibin-lin commented on issue #9636: avoid per-batch blocking in metric
URL: https://github.com/apache/incubator-mxnet/pull/9636#issuecomment-361788582
We should have unit test for this. Please track
This is an automated
cjolivier01 commented on issue #9615: [CMake] Fix OSX double compilation and
add static library
URL: https://github.com/apache/incubator-mxnet/pull/9615#issuecomment-361788507
@de-vri-es Can you please review?
This is an auto
szha opened a new pull request #9636: avoid per-batch blocking in metric
URL: https://github.com/apache/incubator-mxnet/pull/9636
## Description ##
avoid per-batch blocking in metric by moving the blocking logic outside of
for loop.
## Checklist ##
### Essentials ###
- [x] P
marcoabreu commented on a change in pull request #8302: Refactor operators &
MKLDNN
URL: https://github.com/apache/incubator-mxnet/pull/8302#discussion_r164921383
##
File path: prepare_mkldnn.sh
##
@@ -0,0 +1,118 @@
+#!/bin/bash
+
+# Licensed to the Apache Software Foundat
marcoabreu commented on a change in pull request #8302: Refactor operators &
MKLDNN
URL: https://github.com/apache/incubator-mxnet/pull/8302#discussion_r164926483
##
File path: tests/python/gpu/test_gluon_model_zoo.py
##
@@ -0,0 +1,163 @@
+# Licensed to the Apache Software
marcoabreu commented on a change in pull request #8302: Refactor operators &
MKLDNN
URL: https://github.com/apache/incubator-mxnet/pull/8302#discussion_r164925852
##
File path: tests/cpp/operator/batchnorm_test.cc
##
@@ -24,11 +24,14 @@
* \author Chris Olivier
*/
+#if
marcoabreu commented on a change in pull request #8302: Refactor operators &
MKLDNN
URL: https://github.com/apache/incubator-mxnet/pull/8302#discussion_r164926659
##
File path: tests/python/gpu/test_gluon_model_zoo.py
##
@@ -0,0 +1,163 @@
+# Licensed to the Apache Software
nilsjohanbjorck opened a new issue #9635: [Feature request] Division by STD in
mx.io.ImageRecordIter
URL: https://github.com/apache/incubator-mxnet/issues/9635
Dividing input images by a standard deviation is extremely common in CV
applications. Currently, it seems like mx.io.ImageRecordIt
szha commented on a change in pull request #8302: Refactor operators & MKLDNN
URL: https://github.com/apache/incubator-mxnet/pull/8302#discussion_r164924764
##
File path: tests/python/gpu/test_gluon_model_zoo.py
##
@@ -0,0 +1,163 @@
+# Licensed to the Apache Software Founda
szha commented on issue #8461: anchors = mx.symbol.contrib.MultiBoxPrior
URL:
https://github.com/apache/incubator-mxnet/issues/8461#issuecomment-361781563
@apache/mxnet-committers: This issue has been inactive for the past 90 days.
It has no label and needs triage.
For general "how-
szha commented on issue #8486: Something complement to the official MXNet
examples
URL:
https://github.com/apache/incubator-mxnet/issues/8486#issuecomment-361781553
@apache/mxnet-committers: This issue has been inactive for the past 90 days.
It has no label and needs triage.
For ge
szha commented on issue #8487: How to use LSTM or GRU in C++ using
c_predict_api.h?
URL:
https://github.com/apache/incubator-mxnet/issues/8487#issuecomment-361781556
@apache/mxnet-committers: This issue has been inactive for the past 90 days.
It has no label and needs triage.
For g
zheng-da commented on a change in pull request #8302: Refactor operators &
MKLDNN
URL: https://github.com/apache/incubator-mxnet/pull/8302#discussion_r164921462
##
File path: tests/python/gpu/test_gluon_model_zoo.py
##
@@ -0,0 +1,163 @@
+# Licensed to the Apache Software F
szha commented on a change in pull request #8302: Refactor operators & MKLDNN
URL: https://github.com/apache/incubator-mxnet/pull/8302#discussion_r164917612
##
File path: tests/python/gpu/test_gluon_model_zoo.py
##
@@ -0,0 +1,163 @@
+# Licensed to the Apache Software Founda
reminisce commented on a change in pull request #8302: Refactor operators &
MKLDNN
URL: https://github.com/apache/incubator-mxnet/pull/8302#discussion_r164917575
##
File path: src/operator/nn/convolution-inl.h
##
@@ -118,6 +118,29 @@ struct ConvolutionParam : public
dmlc:
cjolivier01 commented on issue #8302: Refactor operators & MKLDNN
URL: https://github.com/apache/incubator-mxnet/pull/8302#issuecomment-361775381
LGTM
This is an automated message from the Apache Git Service.
To respond to the
eric-haibin-lin commented on issue #8302: Refactor operators & MKLDNN
URL: https://github.com/apache/incubator-mxnet/pull/8302#issuecomment-361774412
I'm fine with merging it. What about others?
This is an automated message f
eric-haibin-lin commented on a change in pull request #9625: sparse regression
operators
URL: https://github.com/apache/incubator-mxnet/pull/9625#discussion_r164914751
##
File path: src/operator/regression_output-inl.h
##
@@ -121,6 +174,80 @@ void RegressionBackward(const
eric-haibin-lin commented on a change in pull request #9625: sparse regression
operators
URL: https://github.com/apache/incubator-mxnet/pull/9625#discussion_r164911562
##
File path: src/operator/regression_output-inl.h
##
@@ -121,6 +174,80 @@ void RegressionBackward(const
eric-haibin-lin commented on a change in pull request #9625: sparse regression
operators
URL: https://github.com/apache/incubator-mxnet/pull/9625#discussion_r164910189
##
File path: tests/python/unittest/test_operator.py
##
@@ -216,13 +216,13 @@ def check_slice_channel(dat
eric-haibin-lin commented on a change in pull request #9625: sparse regression
operators
URL: https://github.com/apache/incubator-mxnet/pull/9625#discussion_r164911411
##
File path: src/operator/regression_output-inl.h
##
@@ -95,12 +125,35 @@ void RegressionForward(const n
eric-haibin-lin commented on a change in pull request #9625: sparse regression
operators
URL: https://github.com/apache/incubator-mxnet/pull/9625#discussion_r164910811
##
File path: src/operator/regression_output.cc
##
@@ -24,36 +24,40 @@
#include "./regression_output-i
eric-haibin-lin commented on a change in pull request #9625: sparse regression
operators
URL: https://github.com/apache/incubator-mxnet/pull/9625#discussion_r164911693
##
File path: src/operator/regression_output-inl.h
##
@@ -121,6 +174,80 @@ void RegressionBackward(const
zheng-da commented on a change in pull request #8302: Refactor operators &
MKLDNN
URL: https://github.com/apache/incubator-mxnet/pull/8302#discussion_r164905297
##
File path: src/operator/nn/fully_connected.cc
##
@@ -23,58 +23,165 @@
* \brief fully connect operator
*/
zheng-da commented on a change in pull request #8302: Refactor operators &
MKLDNN
URL: https://github.com/apache/incubator-mxnet/pull/8302#discussion_r164904958
##
File path: Makefile
##
@@ -40,11 +40,11 @@ endif
# use customized config file
include $(config)
-ifeq ($(
zheng-da commented on issue #8302: Refactor operators & MKLDNN
URL: https://github.com/apache/incubator-mxnet/pull/8302#issuecomment-361762157
@piiswrong @eric-haibin-lin @marcoabreu @cjolivier01 @szha @rahul003
@reminisce
Thank you very much for your comments. Do you have more comments
Laurawly commented on issue #8915: NVLink communication pattern updated
URL: https://github.com/apache/incubator-mxnet/pull/8915#issuecomment-361754167
@piiswrong ping.
This is an automated message from the Apache Git Service
asadoughi opened a new issue #9634: [Feature request] Warning message or
exception on mxnet.metric.Loss.update interface for performance
URL: https://github.com/apache/incubator-mxnet/issues/9634
## Description
As shown in the example below, `mxnet.metric.Loss.update` can be over 100
ti
szha commented on a change in pull request #9633: Gluon image-classification
example improvement
URL: https://github.com/apache/incubator-mxnet/pull/9633#discussion_r164881156
##
File path: example/gluon/data.py
##
@@ -49,50 +54,60 @@ def get_cifar10_iterator(batch_size, d
zhreshold opened a new pull request #9633: Gluon image-classification example
improvement
URL: https://github.com/apache/incubator-mxnet/pull/9633
## Description ##
* Example gluon image-classification now able to train imagenet properly.
* More optimizations such as float16 and many
cjolivier01 commented on issue #9609: Enable CPP unit tests in CI
URL: https://github.com/apache/incubator-mxnet/pull/9609#issuecomment-361730772
ok, you can remove the asserts
This is an automated message from the Apache Git
piiswrong opened a new issue #9632: Support pre-kepler GPUs without __shfl_down
instruction
URL: https://github.com/apache/incubator-mxnet/issues/9632
https://discuss.gluon.ai/t/topic/
This is an automated message from th
szha commented on a change in pull request #9614: MobileNetV2
URL: https://github.com/apache/incubator-mxnet/pull/9614#discussion_r164858934
##
File path: python/mxnet/gluon/model_zoo/vision/mobilenet.py
##
@@ -158,3 +166,158 @@ def mobilenet0_25(**kwargs):
The con
szha commented on a change in pull request #9614: MobileNetV2
URL: https://github.com/apache/incubator-mxnet/pull/9614#discussion_r164858839
##
File path: python/mxnet/gluon/model_zoo/vision/mobilenet.py
##
@@ -158,3 +166,158 @@ def mobilenet0_25(**kwargs):
The con
szha commented on a change in pull request #9614: MobileNetV2
URL: https://github.com/apache/incubator-mxnet/pull/9614#discussion_r164859304
##
File path: example/image-classification/symbols/mobilenetv2.py
##
@@ -0,0 +1,184 @@
+# Licensed to the Apache Software Foundation
szha commented on a change in pull request #9614: MobileNetV2
URL: https://github.com/apache/incubator-mxnet/pull/9614#discussion_r164859492
##
File path: example/image-classification/symbols/mobilenetv2.py
##
@@ -0,0 +1,184 @@
+# Licensed to the Apache Software Foundation
zhanghang1989 closed issue #9631: Get Warning When ``#include
``
URL: https://github.com/apache/incubator-mxnet/issues/9631
This is an automated message from the Apache Git Service.
To respond to the message, please log on G
piiswrong commented on issue #9548: updated release version to 1.1.0 on the
mainpage and re-arranged news?
URL: https://github.com/apache/incubator-mxnet/pull/9548#issuecomment-361698494
MXNet model server is actually another project. I wonder whether its
appropriate to advertise it here.
This is an automated email from the ASF dual-hosted git repository.
jxie 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 a0a52b3 Expand gpu-kernel-launch synchro
piiswrong closed pull request #9560: Expand gpu-kernel-launch synchronous error
checking.
URL: https://github.com/apache/incubator-mxnet/pull/9560
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:
A
zhanghang1989 opened a new issue #9631: Get Warning When ``#include
``
URL: https://github.com/apache/incubator-mxnet/issues/9631
When ``#include ``, get the following error:
```bash
/home/ubuntu/git/incubator-mxnet/mshadow/mshadow/././expr_engine-inl.h:438:22:
warning: inline f
larroy commented on issue #9612: CUDNN_STATUS_SUCCESS (4 vs. 0) cuDNN:
CUDNN_STATUS_INTERNAL_ERROR on jetson TX2
URL:
https://github.com/apache/incubator-mxnet/issues/9612#issuecomment-361618152
Hi, there's a fix in mxnet mainline, could you try with examples/ssd instead
of that repositor
eric-haibin-lin commented on issue #9616: Removing a broken tutorial from the
nightly tests
URL: https://github.com/apache/incubator-mxnet/pull/9616#issuecomment-361684891
@aaronmarkham does `test_tutorial.py` work?
This
eric-haibin-lin opened a new pull request #9630: Fix version on website
URL: https://github.com/apache/incubator-mxnet/pull/9630
PR #9535 modified the version of MXNet on the index page which leads to a
broken link. This should only be updated when the voting for the new version
passes. Re
dabraude commented on issue #9274: Is compilation on 32 bit supported?
URL:
https://github.com/apache/incubator-mxnet/issues/9274#issuecomment-361669690
I was having trouble with windows 32bit and I found defining this
-DDMLC_CXX11_THREAD_LOCAL=0
seems to fix it (but it does generate
dabraude commented on issue #9274: Is compilation on 32 bit supported?
URL:
https://github.com/apache/incubator-mxnet/issues/9274#issuecomment-361669690
I was having trouble with windows 32bit and I found defining this
-DDMLC_CXX11_THREAD_LOCAL=0
seems to fix it (but it does generate
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 8205e24 Fix typo in python/mxnet/m
marcoabreu closed pull request #9627: typo
URL: https://github.com/apache/incubator-mxnet/pull/9627
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),
marcoabreu commented on a change in pull request #9614: MobileNetV2
URL: https://github.com/apache/incubator-mxnet/pull/9614#discussion_r164811938
##
File path: example/image-classification/symbols/mobilenetv2.py
##
@@ -0,0 +1,184 @@
+# Licensed to the Apache Software Found
cjolivier01 commented on issue #9545: Profiling discussion
URL:
https://github.com/apache/incubator-mxnet/issues/9545#issuecomment-361647216
Note OpenMP class in src/engine/openmp.cc
```cpp
OpenMP::OpenMP()
: omp_num_threads_set_in_environment(is_env_set("OMP_NUM_THREADS")) {
cjolivier01 commented on issue #9545: Profiling discussion
URL:
https://github.com/apache/incubator-mxnet/issues/9545#issuecomment-361646815
"I think, for mordern deep learning workloads and stencil computation, it
would be better to keep a worker thread pool and bind one worker thread to
cjolivier01 commented on issue #9545: Profiling discussion
URL:
https://github.com/apache/incubator-mxnet/issues/9545#issuecomment-361645985
" if there is no operator parallelism, maybe we should give as many threads
as possible to the running operator, especially for mkldnn operator"
cjolivier01 commented on issue #9545: Profiling discussion
URL:
https://github.com/apache/incubator-mxnet/issues/9545#issuecomment-361645446
"After set USE_OPERATOR_TUNING=0 and rebuilt mxnet, the first 56 threads
were not there anymore. Do you think this behaviour is as expectaton?"
WH
aaronmarkham commented on issue #9616: Removing a broken tutorial from the
nightly tests
URL: https://github.com/apache/incubator-mxnet/pull/9616#issuecomment-361642700
@mbaijal Let's do that! It looks like the tutorials that are tested end with
some kind of `assert` call, but I'm still no
dwSun commented on issue #9614: MobileNetV2
URL: https://github.com/apache/incubator-mxnet/pull/9614#issuecomment-361639938
I have put the MobileNetV2 class in the gluon vision model zoo.
I don't have pretrained models, so this should be modified:
> We provide pre-trained models fo
tornadomeet opened a new issue #9629: do not save gpu memory during fp16
training
URL: https://github.com/apache/incubator-mxnet/issues/9629
hello,
i found when training with fp16, gpu memory is the same as training with
fp32, why not just half gpu memory consume?
marcoabreu commented on issue #9628: kill mxnet using undefault ssh port
URL: https://github.com/apache/incubator-mxnet/pull/9628#issuecomment-361620776
Hello, please follow the template for PRs and provide a description
This
larroy commented on issue #9612: CUDNN_STATUS_SUCCESS (4 vs. 0) cuDNN:
CUDNN_STATUS_INTERNAL_ERROR on jetson TX2
URL:
https://github.com/apache/incubator-mxnet/issues/9612#issuecomment-361618152
Hi, there's a fix in mxnet mainline, could you try with examples/ssd instead
of that repositor
KellenSunderland commented on issue #9609: Enable CPP unit tests in CI
URL: https://github.com/apache/incubator-mxnet/pull/9609#issuecomment-361607662
@cjolivier01 Cpp tests are looking good and ready to go aside from one test
which I think might have some issues on virtualized environments
KellenSunderland commented on issue #9609: Enable CPP unit tests in CI
URL: https://github.com/apache/incubator-mxnet/pull/9609#issuecomment-361607662
@cjolivier01 Cpp tests are looking good and ready to go aside from one test
which I think might have some issues on virtualized environments
1 - 100 of 118 matches
Mail list logo