weixingzhang commented on a change in pull request #8814: csr slice operator,
gpu implementation
URL: https://github.com/apache/incubator-mxnet/pull/8814#discussion_r153413798
##
File path: src/operator/tensor/matrix_op-inl.h
##
@@ -471,10 +467,15 @@ void SliceDimOneCsrImp
eric-haibin-lin commented on issue #8792: Different training performance
between mxnet v. 0.11.0 and v. 0.12.1
URL:
https://github.com/apache/incubator-mxnet/issues/8792#issuecomment-347436732
related issue: https://github.com/apache/incubator-mxnet/issues/7426
---
lx75249 commented on issue #8795: can cpp-package use operators in contrib
directory?
URL:
https://github.com/apache/incubator-mxnet/issues/8795#issuecomment-347436427
No, but it's possible. I just checked the generator. Contrib ops names start
with `_contrib_`, I intentionally filtered a
eric-haibin-lin commented on issue #8658: mxnet random seed does not work for
mx.init.Xavier on both CPU and GPU
URL:
https://github.com/apache/incubator-mxnet/issues/8658#issuecomment-347436372
Good catch! Do you want to post a PR to fix this?
---
eric-haibin-lin commented on issue #8795: can cpp-package use operators in
contrib directory?
URL:
https://github.com/apache/incubator-mxnet/issues/8795#issuecomment-347434998
@lx75249 Does the op generator include contrib ops?
GoodJoey commented on issue #8808: how to change the ssh port when running the
distribute training?
URL:
https://github.com/apache/incubator-mxnet/issues/8808#issuecomment-347434269
thanks guys, @huyangc you are right, adding ":portnumber" works. close this
issue.
--
VGalata commented on issue #8792: Different training performance between mxnet
v. 0.11.0 and v. 0.12.1
URL:
https://github.com/apache/incubator-mxnet/issues/8792#issuecomment-347434365
Hm... It does not seem to be due to the dropout layers. Using version
`0.11.0` gives the same accuracy v
GoodJoey closed issue #8808: how to change the ssh port when running the
distribute training?
URL: https://github.com/apache/incubator-mxnet/issues/8808
This is an automated message from the Apache Git Service.
To respond to
kkk669 commented on issue #8484: MXNet-to-CoreML module: Fixed KeyError for
Reshape
URL: https://github.com/apache/incubator-mxnet/pull/8484#issuecomment-347430542
Why don't you use `_get_attrs()`? It supports both `attr` and `attrs` key.
---
wuterry closed issue #8844: overflow
URL: https://github.com/apache/incubator-mxnet/issues/8844
This is an automated message from the Apache Git Service.
To respond to the message, please log on GitHub and use the
URL above t
wuterry opened a new issue #8844: overflow
URL: https://github.com/apache/incubator-mxnet/issues/8844
Note: Providing complete information in the most concise form is the best
way to get help. This issue template serves as the checklist for essential
information to most of the technical is
reminisce commented on issue #8817: Creating a monodimensional sparse array
returns an error
URL:
https://github.com/apache/incubator-mxnet/issues/8817#issuecomment-347426000
I think it's natural to reshape a vector of shape (10,) to a matrix of shape
(1, 10), placing 10 elements in the s
eric-haibin-lin commented on issue #8817: Creating a monodimensional sparse
array returns an error
URL:
https://github.com/apache/incubator-mxnet/issues/8817#issuecomment-347424436
@reminisce @anirudh2290 @ZiyueHuang @piiswrong any comment on converting
shape (10) to (1,10) explicitly fo
eric-haibin-lin commented on issue #8817: Creating a monodimensional sparse
array returns an error
URL:
https://github.com/apache/incubator-mxnet/issues/8817#issuecomment-347424436
@reminisce @anirudh2290 @ZiyueHuang @piiswrong any comment on converting
shape (10) to (1,10) explicitly fo
reminisce commented on a change in pull request #8839: Fixes memory leaks and
uninitialized variable
URL: https://github.com/apache/incubator-mxnet/pull/8839#discussion_r153402198
##
File path: src/engine/stream_manager.h
##
@@ -117,6 +120,9 @@ RunContext StreamManager::Ge
shivonkar commented on issue #8575: mxnet multicore on LInux in R
URL:
https://github.com/apache/incubator-mxnet/issues/8575#issuecomment-347423440
Hi,
The github is updated
Thank you
Shiv
From: Qiang Kou (KK) [mailto:notificati...@github.com]
Sent: Saturday, November
Soonhwan-Kwon commented on a change in pull request #8787: add CapsNet example
URL: https://github.com/apache/incubator-mxnet/pull/8787#discussion_r153401783
##
File path: example/capsnet/README.md
##
@@ -0,0 +1,49 @@
+**CapsNet-MXNet**
+
Soonhwan-Kwon commented on a change in pull request #8787: add CapsNet example
URL: https://github.com/apache/incubator-mxnet/pull/8787#discussion_r153401954
##
File path: example/capsnet/capsulenet.py
##
@@ -0,0 +1,327 @@
+# Licensed to the Apache Software Foundation (ASF)
Soonhwan-Kwon commented on a change in pull request #8787: add CapsNet example
URL: https://github.com/apache/incubator-mxnet/pull/8787#discussion_r153401783
##
File path: example/capsnet/README.md
##
@@ -0,0 +1,49 @@
+**CapsNet-MXNet**
+
reminisce commented on a change in pull request #7828: New tutorial of
implementing operators in MXNet backend
URL: https://github.com/apache/incubator-mxnet/pull/7828#discussion_r153401429
##
File path: docs/how_to/add_op_in_backend.md
##
@@ -0,0 +1,609 @@
+# A Beginner's
Soonhwan-Kwon commented on issue #8787: add CapsNet example
URL: https://github.com/apache/incubator-mxnet/pull/8787#issuecomment-347421082
@sxjscience Thank you for review,
First we tested on MXNet 0.12.1 and it works well,
and we add 0.392 as option named recon_loss_weight.
In ad
ajinkyaghorpade commented on issue #7044: Error compiling mxnet
URL:
https://github.com/apache/incubator-mxnet/issues/7044#issuecomment-347419283
I'm getting similar issue. @BenEngbers did you manage to resolve the issue?
Thi
weixingzhang commented on issue #8817: Creating a monodimensional sparse array
returns an error
URL:
https://github.com/apache/incubator-mxnet/issues/8817#issuecomment-347418236
@eric-haibin-lin I just made a change #8843 for this issue. Can you help
review it? Thanks!
---
weixingzhang commented on issue #8817: Creating a monodimensional sparse array
returns an error
URL:
https://github.com/apache/incubator-mxnet/issues/8817#issuecomment-347418236
@eric-haibin-lin I just made a change for this issue. Can you help review
it? Thanks! #8843
-
weixingzhang commented on issue #8817: Creating a monodimensional sparse array
returns an error
URL:
https://github.com/apache/incubator-mxnet/issues/8817#issuecomment-347418236
@eric-haibin-lin I just made a change for this issue. Can you help review
it? Thanks! [https://github.com/apach
weixingzhang commented on issue #8817: Creating a monodimensional sparse array
returns an error
URL:
https://github.com/apache/incubator-mxnet/issues/8817#issuecomment-347418236
@eric-haibin-lin I just made a change for this issue. Can you help review
it? Thanks! [https://github.com/apach
weixingzhang opened a new pull request #8843: support monodimensional sparse
array
URL: https://github.com/apache/incubator-mxnet/pull/8843
The end user reported the issue below:
The following code
b1 = nd.sparse.zeros('csr',shape=10, ctx=ctx)
returns
/usr/local/
TaoLv commented on a change in pull request #7828: New tutorial of implementing
operators in MXNet backend
URL: https://github.com/apache/incubator-mxnet/pull/7828#discussion_r153397348
##
File path: docs/how_to/add_op_in_backend.md
##
@@ -0,0 +1,609 @@
+# A Beginner's Gui
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 632c897 Refreshed branch fix_omp_link (#
piiswrong closed pull request #8842: Fix occasional link problem with OpenMP
symbol versioning
URL: https://github.com/apache/incubator-mxnet/pull/8842
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 provenanc
piiswrong closed issue #8262: test_kvstore_gpu.test_row_sparse_pull takes
forever to run and causes build failures
URL: https://github.com/apache/incubator-mxnet/issues/8262
This is an automated message from the Apache Git S
piiswrong closed pull request #8838: fix #8262
URL: https://github.com/apache/incubator-mxnet/pull/8838
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 for
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 ce4b921 fix https://github.com/apache/in
zheng-da commented on a change in pull request #8302: Refactor operators
URL: https://github.com/apache/incubator-mxnet/pull/8302#discussion_r153394021
##
File path: src/operator/nn/cudnn/cudnn_batch_norm.cu
##
@@ -0,0 +1,86 @@
+/*
+ * Licensed to the Apache Software Founda
zheng-da commented on a change in pull request #8302: Refactor operators
URL: https://github.com/apache/incubator-mxnet/pull/8302#discussion_r153394021
##
File path: src/operator/nn/cudnn/cudnn_batch_norm.cu
##
@@ -0,0 +1,86 @@
+/*
+ * Licensed to the Apache Software Founda
YoWhatever commented on issue #6917: ImportError: You used to compile with
protoc --python_out=./ ./caffe.proto
URL:
https://github.com/apache/incubator-mxnet/issues/6917#issuecomment-347408955
@miraclewkf Have U solved that? And how?
--
szha commented on issue #8146: footer overlay covering search results
URL:
https://github.com/apache/incubator-mxnet/issues/8146#issuecomment-347407778
#8841
This is an automated message from the Apache Git Service.
To respon
cjolivier01 opened a new pull request #8842: Fix occasional link problem with
OpenMP symbol versioning
URL: https://github.com/apache/incubator-mxnet/pull/8842
## Description ##
Fix occasional link problem with OpenMP symbol versioning
## Checklist ##
### Essentials ###
- [
szha opened a new pull request #8841: fix search
URL: https://github.com/apache/incubator-mxnet/pull/8841
## Description ##
Fix the problem of footer covering results in search page. See:
http://mxnet-doc.s3-accelerate.dualstack.amazonaws.com/search.html?q=like&check_keywords=yes&area
cjolivier01 commented on a change in pull request #8302: Refactor operators
URL: https://github.com/apache/incubator-mxnet/pull/8302#discussion_r153384405
##
File path: src/operator/nn/cudnn/cudnn_batch_norm.cu
##
@@ -0,0 +1,86 @@
+/*
+ * Licensed to the Apache Software Fou
cjolivier01 commented on a change in pull request #8803: Small first iteration
for profiler: use nonblocking queue
URL: https://github.com/apache/incubator-mxnet/pull/8803#discussion_r153385821
##
File path: src/engine/profiler.cc
##
@@ -56,14 +55,13 @@ Profiler::Profiler(
cjolivier01 commented on a change in pull request #8803: Small first iteration
for profiler: use nonblocking queue
URL: https://github.com/apache/incubator-mxnet/pull/8803#discussion_r153385645
##
File path: src/engine/profiler.cc
##
@@ -113,15 +111,12 @@ OprExecStat *Prof
cjolivier01 commented on a change in pull request #8302: Refactor operators
URL: https://github.com/apache/incubator-mxnet/pull/8302#discussion_r153384405
##
File path: src/operator/nn/cudnn/cudnn_batch_norm.cu
##
@@ -0,0 +1,86 @@
+/*
+ * Licensed to the Apache Software Fou
cjolivier01 commented on a change in pull request #8302: Refactor operators
URL: https://github.com/apache/incubator-mxnet/pull/8302#discussion_r153383825
##
File path: src/operator/nn/cudnn/cudnn_batch_norm.cu
##
@@ -0,0 +1,86 @@
+/*
+ * Licensed to the Apache Software Fou
cjolivier01 commented on a change in pull request #8302: Refactor operators
URL: https://github.com/apache/incubator-mxnet/pull/8302#discussion_r153383825
##
File path: src/operator/nn/cudnn/cudnn_batch_norm.cu
##
@@ -0,0 +1,86 @@
+/*
+ * Licensed to the Apache Software Fou
cjolivier01 commented on a change in pull request #8302: Refactor operators
URL: https://github.com/apache/incubator-mxnet/pull/8302#discussion_r153383825
##
File path: src/operator/nn/cudnn/cudnn_batch_norm.cu
##
@@ -0,0 +1,86 @@
+/*
+ * Licensed to the Apache Software Fou
chenjunweii opened a new issue #8840: [cpp-package] op.h got some error when
compile
URL: https://github.com/apache/incubator-mxnet/issues/8840
i got some error when i compile op.h, there are some codes in "op.h" set
default value like :
`Shape target_shape = [0,0]`
and
zheng-da commented on a change in pull request #8302: Refactor operators
URL: https://github.com/apache/incubator-mxnet/pull/8302#discussion_r153381774
##
File path: src/operator/nn/cudnn/cudnn_batch_norm.cu
##
@@ -0,0 +1,86 @@
+/*
+ * Licensed to the Apache Software Founda
rahul003 commented on issue #8839: Fixes memory leaks and uninitialized variable
URL: https://github.com/apache/incubator-mxnet/pull/8839#issuecomment-347391871
Can someone help me understand why VC++ says more than one default for
switch case?
-
rahul003 commented on issue #8839: Fixes memory leaks and uninitialized variable
URL: https://github.com/apache/incubator-mxnet/pull/8839#issuecomment-347391949
Can someone help me understand why VC++ says more than one default for
switch case? There is only one in stream_manager.h
---
rahul003 commented on issue #8839: Fixes memory leaks and uninitialized variable
URL: https://github.com/apache/incubator-mxnet/pull/8839#issuecomment-347391871
Can someone help me understand why VC++ says more than one default for
switch case?
-
cjolivier01 commented on a change in pull request #8302: Refactor operators
URL: https://github.com/apache/incubator-mxnet/pull/8302#discussion_r15335
##
File path: src/operator/nn/cudnn/cudnn_batch_norm.cu
##
@@ -0,0 +1,86 @@
+/*
+ * Licensed to the Apache Software Fou
cjolivier01 commented on a change in pull request #8302: Refactor operators
URL: https://github.com/apache/incubator-mxnet/pull/8302#discussion_r153377216
##
File path: src/operator/nn/batch_norm.cc
##
@@ -313,45 +314,76 @@ void BatchNormOp::DoBackward(mshadow::Stream *,
cjolivier01 commented on a change in pull request #8302: Refactor operators
URL: https://github.com/apache/incubator-mxnet/pull/8302#discussion_r153377261
##
File path: src/operator/nn/batch_norm.cc
##
@@ -313,45 +314,76 @@ void BatchNormOp::DoBackward(mshadow::Stream *,
cjolivier01 commented on issue #8803: Small first iteration for profiler: use
nonblocking queue
URL: https://github.com/apache/incubator-mxnet/pull/8803#issuecomment-347388615
Ok
On Mon, Nov 27, 2017 at 4:39 PM Rahul Huilgol
wrote:
> *@rahul003* commented on this pull requ
MoritzMaxeiner commented on issue #8836: Backward shape inconsistent with
custom HybridBlock and gluon.loss
URL:
https://github.com/apache/incubator-mxnet/issues/8836#issuecomment-347384351
@reminisce I've removed the time unrolling (and the issue is still being
triggered), but if I remov
MoritzMaxeiner commented on issue #8836: Backward shape inconsistent with
custom HybridBlock and gluon.loss
URL:
https://github.com/apache/incubator-mxnet/issues/8836#issuecomment-347384351
@reminisce I've removed the time unrolling (and the issue is still
triggered), but if I remove eith
MoritzMaxeiner commented on issue #8836: Backward shape inconsistent with
custom HybridBlock and gluon.loss
URL:
https://github.com/apache/incubator-mxnet/issues/8836#issuecomment-347384351
@reminisce I've removed the time unrolling (and the issue is still being
triggered), but if I remov
MoritzMaxeiner commented on issue #8836: Backward shape inconsistent with
custom HybridBlock and gluon.loss
URL:
https://github.com/apache/incubator-mxnet/issues/8836#issuecomment-347384351
@reminisce I've removed the time unrolling, but if I remove either of the
two cells, or the reshape
zheng-da commented on issue #8302: Refactor operators
URL: https://github.com/apache/incubator-mxnet/pull/8302#issuecomment-347380358
I have updated this pull request to address your comments.
I tried `git mv` to move files, but it doesn't work as expected. `git` still
removes and adds f
zheng-da commented on issue #8302: Refactor operators
URL: https://github.com/apache/incubator-mxnet/pull/8302#issuecomment-347380358
I have updated this pull request. I should have addressed almost all
comments.
Do you have any more comments?
If you want to see what files are modifie
zheng-da commented on a change in pull request #8302: Refactor operators
URL: https://github.com/apache/incubator-mxnet/pull/8302#discussion_r153369457
##
File path: src/operator/activation-inl.h
##
@@ -1,198 +0,0 @@
-/*
Review comment:
I tried `git mv` and rebased th
piiswrong closed pull request #8678: [WIP]hue
URL: https://github.com/apache/incubator-mxnet/pull/8678
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
This is an automated email from the ASF dual-hosted git repository.
jxie pushed a commit to branch vision
in repository https://gitbox.apache.org/repos/asf/incubator-mxnet.git
The following commit(s) were added to refs/heads/vision by this push:
new 9cebe62 [WIP]hue (#8678)
9cebe62 is desc
zheng-da commented on a change in pull request #8302: Refactor operators
URL: https://github.com/apache/incubator-mxnet/pull/8302#discussion_r153368427
##
File path: src/operator/nn/batch_norm.cc
##
@@ -313,45 +314,76 @@ void BatchNormOp::DoBackward(mshadow::Stream *,
}
eric-haibin-lin commented on issue #8785: kvstore can not support int64 keys
URL:
https://github.com/apache/incubator-mxnet/issues/8785#issuecomment-347377491
Do you need one key per feature? Or do you have more than 2^32 number of
layers?
eric-haibin-lin commented on issue #8785: kvstore can not support int64 keys
URL:
https://github.com/apache/incubator-mxnet/issues/8785#issuecomment-347377491
Do you need one key per feature? Or do you have more than 2^32 number of
layers?
@rahul003
--
eric-haibin-lin closed issue #8609: git clone command wrong for Scala setup
URL: https://github.com/apache/incubator-mxnet/issues/8609
This is an automated message from the Apache Git Service.
To respond to the message, pleas
eric-haibin-lin commented on issue #8609: git clone command wrong for Scala
setup
URL:
https://github.com/apache/incubator-mxnet/issues/8609#issuecomment-347377232
Should be fixed by #8824
This is an automated message from t
eric-haibin-lin commented on issue #784: Is there a l2 loss layer in mxnet ?
URL: https://github.com/apache/incubator-mxnet/issues/784#issuecomment-347371301
We should update the API doc of `LogisticRegressionOutput ` with what loss
is actually used (binary cross entropy loss).
-
rahul003 opened a new pull request #8839: Coverity fixes
URL: https://github.com/apache/incubator-mxnet/pull/8839
## Description ##
Fixing errors identified by static analysis tool Coverity.
- Fixes memory leaks
- Sets value to uninitialized variable
- Adds a default case to
rahul003 closed pull request #8580: fix some memory leaks, and initializes a
variable
URL: https://github.com/apache/incubator-mxnet/pull/8580
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 th
rahul003 opened a new pull request #8580: fix some memory leaks, and
initializes a variable
URL: https://github.com/apache/incubator-mxnet/pull/8580
## Description ##
Fixing errors identified by static analysis tool Coverity.
1. Fixes memory leaks
2. Sets value to uninitialized
rahul003 commented on a change in pull request #8803: Small first iteration for
profiler: use nonblocking queue
URL: https://github.com/apache/incubator-mxnet/pull/8803#discussion_r153365615
##
File path: src/engine/profiler.cc
##
@@ -113,15 +111,12 @@ OprExecStat *Profile
rahul003 commented on a change in pull request #8803: Small first iteration for
profiler: use nonblocking queue
URL: https://github.com/apache/incubator-mxnet/pull/8803#discussion_r153365321
##
File path: src/engine/profiler.cc
##
@@ -56,14 +55,13 @@ Profiler::Profiler()
szha closed issue #7129: Compile fail on the Jetson TX1
URL: https://github.com/apache/incubator-mxnet/issues/7129
This is an automated message from the Apache Git Service.
To respond to the message, please log on GitHub and
szha closed issue #4272: imagenet example training error
URL: https://github.com/apache/incubator-mxnet/issues/4272
This is an automated message from the Apache Git Service.
To respond to the message, please log on GitHub and
szha closed issue #7642: Max norm regularization
URL: https://github.com/apache/incubator-mxnet/issues/7642
This is an automated message from the Apache Git Service.
To respond to the message, please log on GitHub and use the
szha commented on issue #7642: Max norm regularization
URL:
https://github.com/apache/incubator-mxnet/issues/7642#issuecomment-347373576
This issue is closed due to lack of activity in the last 90 days. Feel free
to ping me to reopen if this is still an active issue. Thanks!
Also, do pl
szha commented on issue #7129: Compile fail on the Jetson TX1
URL:
https://github.com/apache/incubator-mxnet/issues/7129#issuecomment-347373582
This issue is closed due to lack of activity in the last 90 days. Feel free
to ping me to reopen if this is still an active issue. Thanks!
Also
szha commented on issue #4272: imagenet example training error
URL:
https://github.com/apache/incubator-mxnet/issues/4272#issuecomment-347373579
This issue is closed due to lack of activity in the last 90 days. Feel free
to ping me to reopen if this is still an active issue. Thanks!
Als
jeremiedb commented on issue #8792: Different training performance between
mxnet v. 0.11.0 and v. 0.12.1
URL:
https://github.com/apache/incubator-mxnet/issues/8792#issuecomment-347372664
I think the reproducibility issue is solely due to the dropout layer. If you
run a model without dropo
eric-haibin-lin commented on issue #784: Is there a l2 loss layer in mxnet ?
URL: https://github.com/apache/incubator-mxnet/issues/784#issuecomment-347371301
We should update the API doc of `LogisticRegressionOutput ` with what loss
is actually used.
-
brightown opened a new issue #784: Is there a l2 loss layer in mxnet ?
URL: https://github.com/apache/incubator-mxnet/issues/784
Does mxnet support l2 loss now?Like EuclideanLossLayer in caffe?
If I want to add other loss function, e.g. hinge loss,how to implement it?
--
MoritzMaxeiner commented on issue #8836: Backward shape inconsistent with
custom HybridBlock and gluon.loss
URL:
https://github.com/apache/incubator-mxnet/issues/8836#issuecomment-347370079
The issue is also present in 0.12.1.
@reminisce I'll try to reduce it further.
-
cjolivier01 commented on a change in pull request #8302: Refactor operators
URL: https://github.com/apache/incubator-mxnet/pull/8302#discussion_r153356747
##
File path: src/operator/nn/batch_norm.cc
##
@@ -313,45 +314,76 @@ void BatchNormOp::DoBackward(mshadow::Stream *,
eric-haibin-lin commented on issue #8838: fix #8262
URL: https://github.com/apache/incubator-mxnet/pull/8838#issuecomment-347364903
@rahul003 Yes making that explicit improves readability
This is an automated message from the
zheng-da commented on a change in pull request #8302: Refactor operators
URL: https://github.com/apache/incubator-mxnet/pull/8302#discussion_r153355527
##
File path: src/operator/nn/batch_norm.cc
##
@@ -313,45 +314,76 @@ void BatchNormOp::DoBackward(mshadow::Stream *,
}
rahul003 commented on issue #8838: fix #8262
URL: https://github.com/apache/incubator-mxnet/pull/8838#issuecomment-347364012
Looks good.
Although, why remove the wait from CopyFromToImpl and write it three times
now? You wanted it to be explicit?
reminisce commented on issue #8836: Backward shape inconsistent with custom
HybridBlock and gluon.loss
URL:
https://github.com/apache/incubator-mxnet/issues/8836#issuecomment-347363364
I think there is an undefined behavior of backend (C++) code, as the example
behaves differently on diff
reminisce commented on issue #8836: Backward shape inconsistent with custom
HybridBlock and gluon.loss
URL:
https://github.com/apache/incubator-mxnet/issues/8836#issuecomment-347363364
I think there is an undefined behavior of backend (C++) code, as the example
behaves differently on diff
pracheer commented on issue #8782: Caffe to MXNet code translator
URL: https://github.com/apache/incubator-mxnet/pull/8782#issuecomment-347362396
Overall looks good to me.
Few comments:
- Some files don't seem to have any license. E.g.
[batchnorm.st](https://github.com/indhub/mxne
MoritzMaxeiner commented on issue #8836: Backward shape inconsistent with
custom HybridBlock and gluon.loss
URL:
https://github.com/apache/incubator-mxnet/issues/8836#issuecomment-347362013
@reminisce Hm, I've tried out 1.0.0rc0 now and so far I haven't been able to
reproduce the issue in
eric-haibin-lin commented on a change in pull request #8732: rsp push and rsp
pull for comm device, used in kvstore('device')
URL: https://github.com/apache/incubator-mxnet/pull/8732#discussion_r153349168
##
File path: tests/python/gpu/test_kvstore_gpu.py
##
@@ -26,21 +26,
eric-haibin-lin commented on issue #8814: csr slice operator, gpu implementation
URL: https://github.com/apache/incubator-mxnet/pull/8814#issuecomment-347354309
Also ping @anirudh2290 for code review
This is an automated messa
reminisce commented on issue #8836: Backward shape inconsistent with custom
HybridBlock and gluon.loss
URL:
https://github.com/apache/incubator-mxnet/issues/8836#issuecomment-347354180
Seems like some problem of memory invalid access. With the latest master
branch code, the example gives
eric-haibin-lin commented on a change in pull request #8814: csr slice
operator, gpu implementation
URL: https://github.com/apache/incubator-mxnet/pull/8814#discussion_r153324077
##
File path: src/operator/tensor/matrix_op-inl.h
##
@@ -471,10 +467,15 @@ void SliceDimOneCsr
eric-haibin-lin commented on a change in pull request #8732: rsp push and rsp
pull for comm device, used in kvstore('device')
URL: https://github.com/apache/incubator-mxnet/pull/8732#discussion_r153328289
##
File path: src/kvstore/comm.h
##
@@ -619,7 +655,62 @@ class CommD
eric-haibin-lin commented on a change in pull request #8732: rsp push and rsp
pull for comm device, used in kvstore('device')
URL: https://github.com/apache/incubator-mxnet/pull/8732#discussion_r153325170
##
File path: src/kvstore/comm.h
##
@@ -679,8 +770,9 @@ class CommDe
eric-haibin-lin commented on a change in pull request #8732: rsp push and rsp
pull for comm device, used in kvstore('device')
URL: https://github.com/apache/incubator-mxnet/pull/8732#discussion_r153325967
##
File path: src/kvstore/comm.h
##
@@ -526,26 +522,66 @@ class Comm
1 - 100 of 178 matches
Mail list logo