TX2012LH commented on issue #9454: copy tensor from gpu to cpu
URL:
https://github.com/apache/incubator-mxnet/issues/9454#issuecomment-359152702
So, although it works, the pointer `cpu_box1.dptr__` is a random pointer. I
found that it worked after I took value of tensor via
sxjscience closed issue #9506: Using the same seed will generate different
results on different devices
URL: https://github.com/apache/incubator-mxnet/issues/9506
This is an automated message from the Apache Git Service.
To
TX2012LH commented on issue #9454: copy tensor from gpu to cpu
URL:
https://github.com/apache/incubator-mxnet/issues/9454#issuecomment-359152702
I found that it worked after I took value of tensor via `cpu_box1[i][j]`
instead of pointer. The memory of tensor is not continuous.
sxjscience opened a new issue #9506: Using the same seed will generate
different results on different devices
URL: https://github.com/apache/incubator-mxnet/issues/9506
## Description
```python
import mxnet as mx
mx.random.seed(100)
mx.random.normal(0, 1, 10, ctx=mx.gpu(0))
zhangguotai commented on issue #9476: libmxnet.so: undefined symbol:
dorgqr_?
URL:
https://github.com/apache/incubator-mxnet/issues/9476#issuecomment-359147757
I don't know that what relationship of openblas and lapack.
But by $ sudo apt-get install -y libopenblas-dev
pabramowitsch commented on issue #9490: Unable to install mxnet on MacOS .. no
version seemingly compatible with Python3
URL:
https://github.com/apache/incubator-mxnet/issues/9490#issuecomment-359045526
Thank you. I upgraded to high sierra and mxnet installs without a problem
szha commented on issue #9412: Flaky Tests Tracking Issue
URL:
https://github.com/apache/incubator-mxnet/issues/9412#issuecomment-359146133
I fixed #8283 in #9503.
This is an automated message from the Apache Git Service.
To
zhreshold closed pull request #9503: host files needed by caffe converter in s3
URL: https://github.com/apache/incubator-mxnet/pull/9503
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
zhreshold closed issue #8283: Caffe integration test fails because it is not
able to download files from caffe server
URL: https://github.com/apache/incubator-mxnet/issues/8283
This is an automated message from the Apache
This is an automated email from the ASF dual-hosted git repository.
zhreshold 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 1100eca host files needed by caffe
This is an automated email from the ASF dual-hosted git repository.
zhreshold 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 384b736 Prep script and Fixes for
zhreshold closed pull request #9480: Prep script and Fixes for single shot
detector (SSD)
URL: https://github.com/apache/incubator-mxnet/pull/9480
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:
szha commented on issue #9505: [WIP][Review Required] License Fixes based on
Apache RAT Feedback Part 4
URL: https://github.com/apache/incubator-mxnet/pull/9505#issuecomment-359139474
Is this still WIP?
This is an automated
szha commented on issue #9505: [WIP][Review Required] License Fixes based on
Apache RAT Feedback Part 4
URL: https://github.com/apache/incubator-mxnet/pull/9505#issuecomment-359139389
Thanks for fixing it, @mbaijal. The change looks fine and the change won't
affect the packaging
cjolivier01 commented on issue #3946: When predicting, does mxnet provide
thread-safe interface?
URL:
https://github.com/apache/incubator-mxnet/issues/3946#issuecomment-359137708
currently, it will run the ops in parallel, using (possibly) several OMP on
each one independently. OMP
ilibx commented on issue #9489: terminate called after throwing an instance of
'dmlc::Error'
URL:
https://github.com/apache/incubator-mxnet/issues/9489#issuecomment-359137564
@aidan-plenert-macdonald
Thanks for your response, i restart all container, and then run succeeded. i
do not
ilibx closed issue #9489: terminate called after throwing an instance of
'dmlc::Error'
URL: https://github.com/apache/incubator-mxnet/issues/9489
This is an automated message from the Apache Git Service.
To respond to the
eric-haibin-lin commented on issue #3946: When predicting, does mxnet provide
thread-safe interface?
URL:
https://github.com/apache/incubator-mxnet/issues/3946#issuecomment-359136799
> Can mxnet find two independent ops in a computation graph and execute them
parallelly on two cores of
mbaijal opened a new pull request #9505: [WIP][Review Required] License Fixes
based on Apache RAT Feedback Part 4
URL: https://github.com/apache/incubator-mxnet/pull/9505
## Fixing Apache RAT failures - Part 4 ##
PLEASE HELP REVIEW! @szha @pracheer @bhavinthaker
I am not entirely
mbaijal opened a new pull request #9504: [Review Required] License Fixes based
on Apache RAT Feedback Part 3
URL: https://github.com/apache/incubator-mxnet/pull/9504
## Fixing Apache RAT failures - Part 3 ##
PLEASE HELP REVIEW! @indhub @bhavinthaker
Fixing Licenses based on Apache
thinksanky commented on issue #49: fixed windows graphiz download link
URL:
https://github.com/apache/incubator-mxnet-site/pull/49#issuecomment-359134157
@sandeep-krishnamurthy @eric-haibin-lin
Please review and merge this PR. when you get a chance.
thinksanky opened a new pull request #49: fixed windows graphiz download link
URL: https://github.com/apache/incubator-mxnet-site/pull/49
## Description ##
Fixed broken graphiz Windows download link.
This is an
szha opened a new pull request #9503: host files needed by caffe converter
URL: https://github.com/apache/incubator-mxnet/pull/9503
## Description ##
Fixes #8283 by hosting the files by ourselves. Added file integrity check to
avoid repeated download.
## Checklist ##
###
Myrthan opened a new issue #9502: Cannot compile sources
URL: https://github.com/apache/incubator-mxnet/issues/9502
## Description
Cannot fully compile mxnet on ubuntu because of out of memory error.
## Environment info (Required)
Linux ip-172-31-12-26 4.4.0-1048-aws #57-Ubuntu
aidan-plenert-macdonald commented on issue #9461: The fault in the process of
running example
URL:
https://github.com/apache/incubator-mxnet/issues/9461#issuecomment-359128053
This is really hard to read. Can you clean it up and give more detail?
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 23b5657 fix hidden node for batch
szha closed pull request #9487: fix hidden node for batch norm in gluon
api[visualization.py]
URL: https://github.com/apache/incubator-mxnet/pull/9487
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
aidan-plenert-macdonald commented on issue #9489: terminate called after
throwing an instance of 'dmlc::Error'
URL:
https://github.com/apache/incubator-mxnet/issues/9489#issuecomment-359127826
From
mbaijal commented on issue #9046: Fix example
example/reinforcement-learning/a3c
URL: https://github.com/apache/incubator-mxnet/pull/9046#issuecomment-359127024
@eric-haibin-lin @piiswrong
Can you please review this so it can make it into the next release. Thanks!
mbaijal commented on issue #9484: [Review Required] Fixing Licenses: Cleaning
up the Top Level LICENSE file
URL: https://github.com/apache/incubator-mxnet/pull/9484#issuecomment-359126417
@hen Can you please review this change.
aidan-plenert-macdonald opened a new issue #9501: LSTM doesn't work with
Symbols and HybridBlocks
URL: https://github.com/apache/incubator-mxnet/issues/9501
I want to use LSTM with AWS SageMaker, but I am having [problems with
eric-haibin-lin commented on issue #9396: inference speed drop after updating
mxnet from 0.10.0 to 1.0.0
URL:
https://github.com/apache/incubator-mxnet/issues/9396#issuecomment-359124745
@nicklhy what happens if you call `mx.nd.waitall()` before all the forward
calls, remove
aidan-plenert-macdonald commented on issue #7551: How to get the symbol of
gluon's RNN model by hybridizing
URL:
https://github.com/apache/incubator-mxnet/issues/7551#issuecomment-359122807
@piiswrong Is this feature available now?
eric-haibin-lin commented on a change in pull request #9479: add norm operator
for sparse ndarray
URL: https://github.com/apache/incubator-mxnet/pull/9479#discussion_r162756956
##
File path: src/operator/tensor/broadcast_reduce_op.h
##
@@ -816,20 +816,94 @@ struct
mbaijal opened a new pull request #9500: [ReviewRequired] Adding License files
for MXNet Dockerfiles.
URL: https://github.com/apache/incubator-mxnet/pull/9500
## Description ##
Adding a license.md file with the apache 2.0 license to the directories that
contain dockerfiles.
Used a
larroy commented on a change in pull request #9480: Prep script and Fixes for
single shot detector (SSD)
URL: https://github.com/apache/incubator-mxnet/pull/9480#discussion_r162746022
##
File path: src/operator/contrib/multibox_detection.cu
##
@@ -51,7 +51,9 @@ __device__
mbaijal commented on issue #9472: [Review Required] Adding more file formats to
license_header.py (.m, .mk, .R, .cfg)
URL: https://github.com/apache/incubator-mxnet/pull/9472#issuecomment-359094197
Thanks Marco, Kellen, Bhavin for your comments.
I created a new PR for the license
mbaijal closed pull request #9472: [Review Required] Adding more file formats
to license_header.py (.m, .mk, .R, .cfg)
URL: https://github.com/apache/incubator-mxnet/pull/9472
This is a PR merged from a forked repository.
As GitHub hides the original diff on merge, it is displayed below
mbaijal opened a new pull request #9499: [Review Required][UpdatedPR#9472]
Adding more file formats to license_header.py (.m, .mk, .R, .cfg) #9472
URL: https://github.com/apache/incubator-mxnet/pull/9499
## Fixing Apache RAT failures - Part 1##
Updated PR #9472
PLEASE HELP REVIEW!
zhreshold commented on a change in pull request #9480: Prep script and Fixes
for single shot detector (SSD)
URL: https://github.com/apache/incubator-mxnet/pull/9480#discussion_r162734043
##
File path: src/operator/contrib/multibox_detection.cu
##
@@ -51,7 +51,9 @@
marcoabreu commented on issue #8302: Refactor operators & MKLDNN
URL: https://github.com/apache/incubator-mxnet/pull/8302#issuecomment-359078986
Hello, thanks a lot for the benchmarks. Could you please also include
benchmarks with multiple batch-sizes, very few CPU cores (1,2 and 4) and
cjolivier01 opened a new pull request #9498: [DO NOT MERGE] 3p test
URL: https://github.com/apache/incubator-mxnet/pull/9498
## Description ##
(Brief description on what this PR is about)
## Checklist ##
### Essentials ###
- [ ] Passed code style checking (`make lint`)
-
piiswrong commented on issue #9454: copy tensor from gpu to cpu
URL:
https://github.com/apache/incubator-mxnet/issues/9454#issuecomment-359075890
NewTensor doesn't actually allocate memory.
You need to use `new` to allocate memory yourself
piiswrong commented on issue #9474: how to release gpu memory in the garbage
pool?
URL:
https://github.com/apache/incubator-mxnet/issues/9474#issuecomment-359075352
You can set MXNET_GPU_MEM_POOL_RESERVE to a big number (like 80, 90 or 95)
to force mxnet to release memory eagerly. But
piiswrong commented on issue #9483: Creating a symbol variable from a numpy
array
URL:
https://github.com/apache/incubator-mxnet/issues/9483#issuecomment-359074759
You can use set_params after initialization.
This is an
cjolivier01 commented on a change in pull request #9373: Proof of concept for
exception handling in operators (DON'T MERGE)
URL: https://github.com/apache/incubator-mxnet/pull/9373#discussion_r162719578
##
File path: src/engine/threaded_engine.cc
##
@@ -391,18 +394,29 @@
cjolivier01 commented on a change in pull request #9373: Proof of concept for
exception handling in operators (DON'T MERGE)
URL: https://github.com/apache/incubator-mxnet/pull/9373#discussion_r162719371
##
File path: src/engine/threaded_engine.cc
##
@@ -391,18 +394,29 @@
piiswrong commented on issue #9319: repeat run constructor function of
SymbolBlock when using gluon interface
URL:
https://github.com/apache/incubator-mxnet/issues/9319#issuecomment-359074299
Operators was never designed to store persistent states. They should be
stored as parameter or
This is an automated email from the ASF dual-hosted git repository.
jxie pushed a change to branch dev
in repository https://gitbox.apache.org/repos/asf/incubator-mxnet.git.
was ce87e65 Update tests for log2, log10 (#8271)
The revisions that were on this branch are still contained in
This is an automated email from the ASF dual-hosted git repository.
jxie pushed a change to branch mli-patch-1
in repository https://gitbox.apache.org/repos/asf/incubator-mxnet.git.
was d4153cc Reshuffle data for each epoch in NDArrayIter
The revisions that were on this branch are still
cjolivier01 commented on issue #8972: Profiling enhancements, python API, vtune
and chrome tracing objects, etc.
URL: https://github.com/apache/incubator-mxnet/pull/8972#issuecomment-359062516
@thirdwing Can you please review this R commit?
cjolivier01 commented on issue #8972: Profiling enhancements, python API, vtune
and chrome tracing objects, etc.
URL: https://github.com/apache/incubator-mxnet/pull/8972#issuecomment-359061152
@sergeykolychev Can you please check Perl commit (I am almost certain it
needs changes, but I am
piiswrong commented on issue #9495: bugfix for kAddTo in swap_axis
URL: https://github.com/apache/incubator-mxnet/pull/9495#issuecomment-359054204
Temp space is not necessay. Change line
https://github.com/apache/incubator-mxnet/pull/9495/files#diff-f299af08ae07ceaecbd468970ef7ee9cR128
to
sxjscience commented on a change in pull request #9495: bugfix for kAddTo in
swap_axis
URL: https://github.com/apache/incubator-mxnet/pull/9495#discussion_r162699288
##
File path: src/operator/swapaxis-inl.h
##
@@ -218,6 +227,11 @@ class SwapAxisProp : public
zanbri commented on issue #4009: numpy indexing and symbolic shapes
URL:
https://github.com/apache/incubator-mxnet/issues/4009#issuecomment-359048511
I'd like to reopen this issue as I still don't see a general solution for
how to slice a symbol to the shape of another symbol. Previously,
pabramowitsch commented on issue #9490: Unable to install mxnet on MacOS .. no
version seemingly compatible with Python3
URL:
https://github.com/apache/incubator-mxnet/issues/9490#issuecomment-359045526
Thank you. I was wondering about that, but as python3 itself didn't
complain about
cjolivier01 commented on a change in pull request #9479: add norm operator for
sparse ndarray
URL: https://github.com/apache/incubator-mxnet/pull/9479#discussion_r162692758
##
File path: src/operator/tensor/broadcast_reduce_op.h
##
@@ -821,20 +821,93 @@ struct ReduceGrad
sxjscience commented on issue #9493: Data Loading API for NDArray
URL:
https://github.com/apache/incubator-mxnet/issues/9493#issuecomment-359043215
@piiswrong @sha Would the new `gluon.data.RecordFileDataset` satisfy the
need?
sxjscience commented on issue #9493: Data Loading API for NDArray
URL:
https://github.com/apache/incubator-mxnet/issues/9493#issuecomment-359043215
@piiswrong @sha Would the new `gluon.data.RecordFileDataset` satisfy the
need?
szha commented on issue #9490: Unable to install mxnet on MacOS .. no version
seemingly compatible with Python3
URL:
https://github.com/apache/incubator-mxnet/issues/9490#issuecomment-359041976
Hi @pabramowitsch. I just realized that your kernel version 13.4.0 indicates
you're using OSX
pabramowitsch commented on issue #9490: Unable to install mxnet on MacOS .. no
version seemingly compatible with Python3
URL:
https://github.com/apache/incubator-mxnet/issues/9490#issuecomment-359037096
Thanks szha! Yes, I've tried not specifying any version as you indicated,
or tried
sxjscience closed issue #8328: nd.random_normal segfaults if shape not provided
URL: https://github.com/apache/incubator-mxnet/issues/8328
This is an automated message from the Apache Git Service.
To respond to the message,
sxjscience closed issue #9491: is_train in model.backward will change the bias
parameter?
URL: https://github.com/apache/incubator-mxnet/issues/9491
This is an automated message from the Apache Git Service.
To respond to
sxjscience commented on issue #9495: bugfix for kAddTo in swap_axis
URL: https://github.com/apache/incubator-mxnet/pull/9495#issuecomment-359014216
We need to test the case ?grad_req=addto?. There seems to be many untested
OPs. Here is an example of testing the correctness of AddTo:
larroy commented on issue #9480: Prep script and Fixes for single shot detector
(SSD)
URL: https://github.com/apache/incubator-mxnet/pull/9480#issuecomment-358996388
@zhreshold
This is an automated message from the Apache
larroy commented on a change in pull request #9480: Prep script and Fixes for
single shot detector (SSD)
URL: https://github.com/apache/incubator-mxnet/pull/9480#discussion_r162645590
##
File path: example/ssd/data/demo/download_demo_images.py
##
@@ -1,3 +1,6 @@
tdomhan commented on issue #9495: bugfix for kAddTo in swap_axis
URL: https://github.com/apache/incubator-mxnet/pull/9495#issuecomment-358981642
thanks! I can confirm that this fixes the swapaxes issue observed in Sockeye
PR https://github.com/awslabs/sockeye/pull/272
asmushetzel commented on issue #9497: Ensure that all operators properly
support kAddTo in backward pass
URL:
https://github.com/apache/incubator-mxnet/issues/9497#issuecomment-358981556
@KellenSunderland @marcoabreu @cjolivier01
asmushetzel opened a new issue #9497: Ensure that all operators properly
support kAddTo in backward pass
URL: https://github.com/apache/incubator-mxnet/issues/9497
PR9495 fixes one operation where the backward pass of an operator did not
obey the "req" argument properly (resp. not at all)
asmushetzel commented on issue #9495: bugfix for kAddTo in swap_axis
URL: https://github.com/apache/incubator-mxnet/pull/9495#issuecomment-358978431
@KellenSunderland @fhieber @tdomhan
This is an automated message from the
asmushetzel opened a new pull request #9495: bugfix for kAddTo in swap_axis
URL: https://github.com/apache/incubator-mxnet/pull/9495
## Description ##
the operator switch_axis was ignoring the "req" argument. That leads to
wrong gradients being computed if an operator fans out to two or
ZiyueHuang commented on a change in pull request #9479: add norm operator for
sparse ndarray
URL: https://github.com/apache/incubator-mxnet/pull/9479#discussion_r162623029
##
File path: src/operator/tensor/broadcast_reduce_op.h
##
@@ -816,20 +816,94 @@ struct ReduceGrad
KellenSunderland commented on a change in pull request #9480: Prep script and
Fixes for single shot detector (SSD)
URL: https://github.com/apache/incubator-mxnet/pull/9480#discussion_r162620927
##
File path: example/ssd/data/demo/download_demo_images.py
##
@@ -1,3 +1,6 @@
asmushetzel commented on issue #9306: Sequence Operator Improvements
URL: https://github.com/apache/incubator-mxnet/pull/9306#issuecomment-358958669
I just accidentally stumbled upon the bug with the kAddTo in SequenceMask
(on a slightly older codebase). Nice to see it already fixed ;-)
szha commented on issue #4045: inferred shape error with FullyConnected layer
URL:
https://github.com/apache/incubator-mxnet/issues/4045#issuecomment-358953295
@apache/mxnet-committers: This issue has been inactive for the past 90 days.
It has no label and needs triage.
For general
szha commented on issue #8328: nd.random_normal segfaults if shape not provided
URL:
https://github.com/apache/incubator-mxnet/issues/8328#issuecomment-358953320
@apache/mxnet-committers: This issue has been inactive for the past 90 days.
It has no label and needs triage.
For
szha commented on issue #8355: Wrong argument in mpi tracker
URL:
https://github.com/apache/incubator-mxnet/issues/8355#issuecomment-358953312
@apache/mxnet-committers: This issue has been inactive for the past 90 days.
It has no label and needs triage.
For general "how-to"
szha commented on issue #8341: Training error always fluctuates and doesn't
decrease.
URL:
https://github.com/apache/incubator-mxnet/issues/8341#issuecomment-358953303
@apache/mxnet-committers: This issue has been inactive for the past 90 days.
It has no label and needs triage.
KellenSunderland closed pull request #9469: WIP: Fp16 sockeye - MXNet required
changes.
URL: https://github.com/apache/incubator-mxnet/pull/9469
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
larroy commented on issue #9476: libmxnet.so: undefined symbol:
dorgqr_?
URL:
https://github.com/apache/incubator-mxnet/issues/9476#issuecomment-358940958
can you try to disable LAPACK? Do you need LAPACK?
Did you compile lapack and openblas yourself?
larroy commented on issue #9397: [CMake] Remove linker flags from CXX_FLAGS
when using OPENCV. Exclude binaries from lint. Fix im2rec on cuda.
URL: https://github.com/apache/incubator-mxnet/pull/9397#issuecomment-358940355
It's explained in the body of the PR: **(clang: warning:
pengzhao-intel commented on issue #8302: Refactor operators & MKLDNN
URL: https://github.com/apache/incubator-mxnet/pull/8302#issuecomment-358939263
Regarding the inference performance of **Batchsize=1**, the new MKL-DNN
shows better performance than MKLML engine as below table.
*I
larroy commented on issue #9397: [CMake] Remove linker flags from CXX_FLAGS
when using OPENCV. Exclude binaries from lint. Fix im2rec on cuda.
URL: https://github.com/apache/incubator-mxnet/pull/9397#issuecomment-358940355
It's explained in the body of the PR: **(clang: warning:
larroy commented on issue #8880: correct usage of bool arguments from command
line
URL: https://github.com/apache/incubator-mxnet/pull/8880#issuecomment-358940025
@sunyonggang @piiswrong my change fixes the doc:
https://github.com/apache/incubator-mxnet/pull/9457
KellenSunderland opened a new pull request #9469: WIP: Fp16 sockeye - MXNet
required changes.
URL: https://github.com/apache/incubator-mxnet/pull/9469
## Description ##
Testing some MXNet changes that would be required in order to support fp16
NMT. Please don't merge, will require a
pengzhao-intel commented on issue #8302: Refactor operators & MKLDNN
URL: https://github.com/apache/incubator-mxnet/pull/8302#issuecomment-358939263
Regarding the inference performance of **Batchsize=1**, the new MKL-DNN
shows better performance than MKLML engine as below table.
sunyonggang commented on issue #8880: correct usage of bool arguments from
command line
URL: https://github.com/apache/incubator-mxnet/pull/8880#issuecomment-358930304
Oh?, it's my first time to use im2rec.py to get like this:
```
python ~/incubator-mxnet/tools/im2rec.py --list True
GSanchis commented on issue #8669: module.Module and CSVIter
URL:
https://github.com/apache/incubator-mxnet/issues/8669#issuecomment-358917324
After quite some time stumbling into this problem, apparently at random, I
believe I have found out what the problem was, after reading
Rikorose opened a new issue #9493: Data Loading API for NDArray
URL: https://github.com/apache/incubator-mxnet/issues/9493
## Description
I am struggling to find an efficient way of loading non image data that has
the shape of `(f, t)`, where `f` are my number of features and `t`
zheng-da commented on issue #8302: Refactor operators & MKLDNN
URL: https://github.com/apache/incubator-mxnet/pull/8302#issuecomment-358894035
@pengzhao-intel could you also put up your evaluation on the speed of this
MKLDNN integration?
90 matches
Mail list logo