[incubator-mxnet-site] branch asf-site updated: Bump the publish timestamp.
This is an automated email from the ASF dual-hosted git repository. aaronmarkham pushed a commit to branch asf-site in repository https://gitbox.apache.org/repos/asf/incubator-mxnet-site.git The following commit(s) were added to refs/heads/asf-site by this push: new 7d40fc1 Bump the publish timestamp. 7d40fc1 is described below commit 7d40fc16522c54c643b0a19ccb40d8d98263ada5 Author: mxnet-ci AuthorDate: Sat Sep 21 06:40:56 2019 + Bump the publish timestamp. --- date.txt | 1 + 1 file changed, 1 insertion(+) diff --git a/date.txt b/date.txt new file mode 100644 index 000..abd7ab5 --- /dev/null +++ b/date.txt @@ -0,0 +1 @@ +Sat Sep 21 06:40:56 UTC 2019
[GitHub] [incubator-mxnet] hzfan commented on a change in pull request #15855: [Numpy] cumprod
hzfan commented on a change in pull request #15855: [Numpy] cumprod
URL: https://github.com/apache/incubator-mxnet/pull/15855#discussion_r326845669
##
File path: ci/docker/runtime_functions.sh
##
@@ -367,6 +367,7 @@ build_ubuntu_cpu_openblas() {
set -ex
export CC="gcc"
export CXX="g++"
+pip3 install --user psutil
Review comment:
> How about you add Ubuntu tvm to the CPU dockerfile?
Yes, but I will also need a new file `ubuntu_tvm_cpu.sh` because
`ubuntu_tvm.sh` installs tvm with cuda and cannot be run on cpu.
> I mean if we want to run tvm cpu tests, we should also install it, right?
No, it is not installed in ci. TVM is a 3rdparty in MXNet, so it is
installed in cmake.
Will a new file like `ubuntu_tvm_cpu.sh` be acceptable?
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org
With regards,
Apache Git Services
[GitHub] [incubator-mxnet] JiaoPaner edited a comment on issue #15632: Building MxNet with CPP_PACKAGE on Windows10 (2019-07-23)
JiaoPaner edited a comment on issue #15632: Building MxNet with CPP_PACKAGE on Windows10 (2019-07-23) URL: https://github.com/apache/incubator-mxnet/issues/15632#issuecomment-533443667 @QueensGambit I read the #15144 ,modified the line " #if !defined(_MSC_VER)" which is 744 in "include/mxnet/tuple.h" to "#if !(defined(_MSC_VER) && _MSC_VER < 1900)" .Then, compile again that config includes USE_CPP_PACKAGE, my operating system is win10 x64 and IDE is Visual Studio 2015. After that the compile result has no any errors and no any above errors. And the examples can run without error , the solution does work. This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-mxnet] WenmuZhou opened a new issue #16232: how can I load a pre-trained model with different shape and same name of last layer
WenmuZhou opened a new issue #16232: how can I load a pre-trained model with different shape and same name of last layer URL: https://github.com/apache/incubator-mxnet/issues/16232 This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-mxnet] WenmuZhou commented on issue #15797: how can I concat two dataset with a sepcial ratio?
WenmuZhou commented on issue #15797: how can I concat two dataset with a sepcial ratio? URL: https://github.com/apache/incubator-mxnet/issues/15797#issuecomment-533758392 I have write a dataset like this and it work https://github.com/WenmuZhou/crnn.gluon/blob/master/data_loader/dataset.py#L140-L195 This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-mxnet] gyshi commented on a change in pull request #16025: Numpy add numpy op left_shift and right_shift
gyshi commented on a change in pull request #16025: Numpy add numpy op
left_shift and right_shift
URL: https://github.com/apache/incubator-mxnet/pull/16025#discussion_r326840977
##
File path: src/operator/numpy/np_elemwise_broadcast_op.cc
##
@@ -144,5 +165,21 @@
MXNET_OPERATOR_REGISTER_NP_BINARY_SCALAR(_backward_npi_rcopysign_scalar)
.set_attr("FCompute",
BinaryScalarOp::Backward);
+MXNET_OPERATOR_REGISTER_NP_BINARY_SCALAR(_npi_left_shift_scalar)
+.set_attr("FCompute", BitScalarCompute)
+.set_attr("FGradient", MakeZeroGradNodes);
Review comment:
it's right, i wll add backward
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org
With regards,
Apache Git Services
[incubator-mxnet-site] branch asf-site updated: Bump the publish timestamp.
This is an automated email from the ASF dual-hosted git repository. aaronmarkham pushed a commit to branch asf-site in repository https://gitbox.apache.org/repos/asf/incubator-mxnet-site.git The following commit(s) were added to refs/heads/asf-site by this push: new f748aec Bump the publish timestamp. f748aec is described below commit f748aec79f50d800c76824ef060f9f9e1c14695d Author: mxnet-ci AuthorDate: Sat Sep 21 00:41:08 2019 + Bump the publish timestamp. --- date.txt | 1 + 1 file changed, 1 insertion(+) diff --git a/date.txt b/date.txt new file mode 100644 index 000..55e4300 --- /dev/null +++ b/date.txt @@ -0,0 +1 @@ +Sat Sep 21 00:41:08 UTC 2019
[GitHub] [incubator-mxnet] ChaiBapchya commented on issue #16174: How to build new operators?
ChaiBapchya commented on issue #16174: How to build new operators? URL: https://github.com/apache/incubator-mxnet/issues/16174#issuecomment-533746591 With today's new updated website launch - https://mxnet.incubator.apache.org/api/faq/new_op should be the link @zachgk This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[incubator-mxnet] branch master updated (618c481 -> be7296b)
This is an automated email from the ASF dual-hosted git repository. apeforest pushed a change to branch master in repository https://gitbox.apache.org/repos/asf/incubator-mxnet.git. from 618c481 [MXNET-1422] Fix wrong results of min([inf, inf]) and max([-inf,-inf]) (#16226) add be7296b removing MXNDArrayLoadFromBuffer64 and MXNDArrayLoad64 (#16203) No new revisions were added by this update. Summary of changes: include/mxnet/c_api.h | 13 - 1 file changed, 13 deletions(-)
[GitHub] [incubator-mxnet] apeforest merged pull request #16203: removing 64 bit un-used APIs MXNDArrayLoadFromBuffer64 and MXNDArrayLoad64
apeforest merged pull request #16203: removing 64 bit un-used APIs MXNDArrayLoadFromBuffer64 and MXNDArrayLoad64 URL: https://github.com/apache/incubator-mxnet/pull/16203 This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-mxnet] ptrendx commented on issue #15657: Eliminate common expressions
ptrendx commented on issue #15657: Eliminate common expressions URL: https://github.com/apache/incubator-mxnet/pull/15657#issuecomment-533742273 @DickJC123 Ok, answering your questions: 1. This definition of Node equality is quite tied to what I want to do with them, so I believe, at least right now, it belongs to CSE code. 2. While I agree it would be beneficial, it would also be pretty hard to do - the parameter structure is kept without the type information as the `shared_ptr`. Comparing memory (which could be dine if `any` gave size of the type held) is also not really feasible since `dict` is actually part of the `Parameter` structure, so that would be different. 3. This is the purpose of this `THasDeterministicOutput` property. 4. I will think about that. This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-mxnet] ChaiBapchya commented on issue #16231: [Website] MXNet Version selection Drop-down makes page jump to top of page
ChaiBapchya commented on issue #16231: [Website] MXNet Version selection Drop-down makes page jump to top of page URL: https://github.com/apache/incubator-mxnet/issues/16231#issuecomment-533742086 @mxnet-label-bot add [website-beta] @aaronmarkham What do you think? This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-mxnet] ChaiBapchya opened a new issue #16231: [Website] MXNet Version selection Drop-down makes page jump to top of page
ChaiBapchya opened a new issue #16231: [Website] MXNet Version selection Drop-down makes page jump to top of page URL: https://github.com/apache/incubator-mxnet/issues/16231 Upon selecting the version from the drop-down, it moves the screen to the top of the page - not the desired outcome.  However, upon selecting the rest of the options - OS Platform, Langugage binding, GPU/CPU etc the screen remains in the original position (instead of moving to the top of the page) which is the desired outcome. I'm guessing it involves playing with the button's on-click and using `$e.preventDefault();` somewhere around showContent() after populating new options https://stackoverflow.com/questions/43611173/button-onclick-makes-page-jump-to-top-of-page This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-mxnet] mxnet-label-bot commented on issue #16231: [Website] MXNet Version selection Drop-down makes page jump to top of page
mxnet-label-bot commented on issue #16231: [Website] MXNet Version selection Drop-down makes page jump to top of page URL: https://github.com/apache/incubator-mxnet/issues/16231#issuecomment-533741984 Hey, this is the MXNet Label Bot. Thank you for submitting the issue! I will try and suggest some labels so that the appropriate MXNet community members can help resolve it. Here are my recommended label(s): Doc This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-mxnet] salmanmashayekh opened a new issue #16230: Loading Sagemaker NTM Artifacts
salmanmashayekh opened a new issue #16230: Loading Sagemaker NTM Artifacts
URL: https://github.com/apache/incubator-mxnet/issues/16230
I have trained a Neural Topic Model with Sagemaker and now I am trying to
load/deploy the model locally. The artifacts include a `symbol` and an a
`parameters` file.
I am using the following to load the model:
```
sym, arg_params, aux_params = mx.model.load_checkpoint(prefix, iteration)
module_model = mx.mod.Module(symbol=sym, label_names=None, context=mx.cpu())
```
But when I try to `bind` the model:
```
module_model.bind(
for_training = False,
data_shapes = [('data', (1, VOCAB_SIZE))]
)
```
It fails with the following error:
```
---
MXNetErrorTraceback (most recent call last)
~/anaconda3/envs/python3/lib/python3.6/site-packages/mxnet/symbol/symbol.py
in simple_bind(self, ctx, grad_req, type_dict, stype_dict, group2ctx,
shared_arg_names, shared_exec, shared_buffer, **kwargs)
1622
shared_exec_handle,
-> 1623
ctypes.byref(exe_handle)))
1624 except MXNetError as e:
~/anaconda3/envs/python3/lib/python3.6/site-packages/mxnet/base.py in
check_call(ret)
252 if ret != 0:
--> 253 raise MXNetError(py_str(_LIB.MXGetLastError()))
254
MXNetError: Error in operator sample_normal0: vector::_M_range_insert
During handling of the above exception, another exception occurred:
RuntimeError Traceback (most recent call last)
in ()
8 for_training = True,
9 data_shapes = [('data', (1,VOCAB_SIZE))],
---> 10 force_rebind = True,
11 )
12
~/anaconda3/envs/python3/lib/python3.6/site-packages/mxnet/module/module.py
in bind(self, data_shapes, label_shapes, for_training, inputs_need_grad,
force_rebind, shared_module, grad_req)
427
fixed_param_names=self._fixed_param_names,
428
grad_req=grad_req, group2ctxs=self._group2ctxs,
--> 429
state_names=self._state_names)
430 self._total_exec_bytes = self._exec_group._total_exec_bytes
431 if shared_module is not None:
~/anaconda3/envs/python3/lib/python3.6/site-packages/mxnet/module/executor_group.py
in __init__(self, symbol, contexts, workload, data_shapes, label_shapes,
param_names, for_training, inputs_need_grad, shared_group, logger,
fixed_param_names, grad_req, state_names, group2ctxs)
277 self.num_outputs = len(self.symbol.list_outputs())
278
--> 279 self.bind_exec(data_shapes, label_shapes, shared_group)
280
281 def decide_slices(self, data_shapes):
~/anaconda3/envs/python3/lib/python3.6/site-packages/mxnet/module/executor_group.py
in bind_exec(self, data_shapes, label_shapes, shared_group, reshape)
373 else:
374 self.execs.append(self._bind_ith_exec(i,
data_shapes_i, label_shapes_i,
--> 375 shared_group))
376
377 self.data_shapes = data_shapes
~/anaconda3/envs/python3/lib/python3.6/site-packages/mxnet/module/executor_group.py
in _bind_ith_exec(self, i, data_shapes, label_shapes, shared_group)
660type_dict=input_types,
shared_arg_names=self.param_names,
661shared_exec=shared_exec,
group2ctx=group2ctx,
--> 662
shared_buffer=shared_data_arrays, **input_shapes)
663 self._total_exec_bytes +=
int(executor.debug_str().split('\n')[-3].split()[1])
664 return executor
~/anaconda3/envs/python3/lib/python3.6/site-packages/mxnet/symbol/symbol.py
in simple_bind(self, ctx, grad_req, type_dict, stype_dict, group2ctx,
shared_arg_names, shared_exec, shared_buffer, **kwargs)
1627 error_msg += "%s: %s\n" % (k, v)
1628 error_msg += "%s" % e
-> 1629 raise RuntimeError(error_msg)
1630
1631 # update shared_buffer
RuntimeError: simple_bind error. Arguments:
data: (1, 52908)
Error in operator sample_normal0: vector::_M_range_insert
```
From the model architecture (https://arxiv.org/pdf/1809.02687.pdf), I know
that the input data shape is a vector with `VOCAB_SIZE` length.
Any ideas what I am doing wrong?
This is an automated message from the Apache Git Service.
[GitHub] [incubator-mxnet] rondogency opened a new pull request #16228: [DO NOT MERGE] Debugging UNIX_GPU CI time regression since 15164
rondogency opened a new pull request #16228: [DO NOT MERGE] Debugging UNIX_GPU CI time regression since 15164 URL: https://github.com/apache/incubator-mxnet/pull/16228 ## Description ## (Brief description on what this PR is about) ## Checklist ## ### Essentials ### Please feel free to remove inapplicable items for your PR. - [ ] The PR title starts with [MXNET-$JIRA_ID], where $JIRA_ID refers to the relevant [JIRA issue](https://issues.apache.org/jira/projects/MXNET/issues) created (except PRs with tiny changes) - [ ] Changes are complete (i.e. I finished coding on this PR) - [ ] All changes have test coverage: - Unit tests are added for small changes to verify correctness (e.g. adding a new operator) - Nightly tests are added for complicated/long-running ones (e.g. changing distributed kvstore) - Build tests will be added for build configuration changes (e.g. adding a new build option with NCCL) - [ ] Code is well-documented: - For user-facing API changes, API doc string has been updated. - For new C++ functions in header files, their functionalities and arguments are documented. - For new examples, README.md is added to explain the what the example does, the source of the dataset, expected performance on test set and reference to the original paper if applicable - Check the API doc at http://mxnet-ci-doc.s3-accelerate.dualstack.amazonaws.com/PR-$PR_ID/$BUILD_ID/index.html - [ ] To the my best knowledge, examples are either not affected by this change, or have been fixed to be compatible with this change ### Changes ### - [ ] Feature1, tests, (and when applicable, API doc) - [ ] Feature2, tests, (and when applicable, API doc) ## Comments ## - If this change is a backward incompatible change, why must this change be made. - Interesting edge cases to note here This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-mxnet] dtracz opened a new pull request #16229: Pr transpose
dtracz opened a new pull request #16229: Pr transpose URL: https://github.com/apache/incubator-mxnet/pull/16229 Fast pseudo-2D transpose kernel. Supports only transposes that satisfy: Exists n and m such that: params = (0, ..., n-1, n+m, ..., params.size, n, ..., n+m-1) Example: (0, 2, 3, 1) or (0, 3, 1, 2), but not (0, 2, 1, 3). This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-mxnet] wkcn closed issue #16206: Wrong results of min([inf, inf]) and max([-inf, -inf])
wkcn closed issue #16206: Wrong results of min([inf, inf]) and max([-inf,-inf]) URL: https://github.com/apache/incubator-mxnet/issues/16206 This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-mxnet] wkcn commented on issue #16206: Wrong results of min([inf, inf]) and max([-inf, -inf])
wkcn commented on issue #16206: Wrong results of min([inf, inf]) and max([-inf,-inf]) URL: https://github.com/apache/incubator-mxnet/issues/16206#issuecomment-533734755 Close it since the issue has been addressed in PR #16226 This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[incubator-mxnet] branch master updated (d61ed3f -> 618c481)
This is an automated email from the ASF dual-hosted git repository. marcoabreu pushed a change to branch master in repository https://gitbox.apache.org/repos/asf/incubator-mxnet.git. from d61ed3f Solve #14116, #15143 (#15144) add 618c481 [MXNET-1422] Fix wrong results of min([inf, inf]) and max([-inf,-inf]) (#16226) No new revisions were added by this update. Summary of changes: 3rdparty/mshadow/mshadow/base.h| 43 -- tests/python/unittest/test_operator.py | 17 ++ 2 files changed, 58 insertions(+), 2 deletions(-)
[GitHub] [incubator-mxnet] marcoabreu commented on issue #16154: Deprecated MXNET website is still online ?
marcoabreu commented on issue #16154: Deprecated MXNET website is still online ? URL: https://github.com/apache/incubator-mxnet/issues/16154#issuecomment-533731855 @ThomasDelteil This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-mxnet] marcoabreu merged pull request #16226: [MXNET-1422] Fix wrong results of min([inf, inf]) and max([-inf, -inf])
marcoabreu merged pull request #16226: [MXNET-1422] Fix wrong results of min([inf, inf]) and max([-inf,-inf]) URL: https://github.com/apache/incubator-mxnet/pull/16226 This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-mxnet] anirudhacharya commented on issue #10051: Nondeterministic order of data in a data batch from NDArrayIter
anirudhacharya commented on issue #10051: Nondeterministic order of data in a data batch from NDArrayIter URL: https://github.com/apache/incubator-mxnet/issues/10051#issuecomment-533728736 @eric-haibin-lin can this issue be closed? This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-mxnet] zhreshold commented on issue #13593: Low CPU usage of MXNet in subprocesses
zhreshold commented on issue #13593: Low CPU usage of MXNet in subprocesses URL: https://github.com/apache/incubator-mxnet/issues/13593#issuecomment-533722252 @YutingZhang Just tested out the master version, the ENV variable `OMP_NUM_THREADS` can now effectively control the OMP threads each worker is allowed to use. For example, `OMP_NUM_THREADS=32 python3 mxnet_cpu_test.py --num-workers=2` gives  This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[incubator-mxnet] branch master updated (986cecd -> d61ed3f)
This is an automated email from the ASF dual-hosted git repository. zachgk pushed a change to branch master in repository https://gitbox.apache.org/repos/asf/incubator-mxnet.git. from 986cecd Update MKL-DNN dependency (#16073) add d61ed3f Solve #14116, #15143 (#15144) No new revisions were added by this update. Summary of changes: include/mxnet/tuple.h | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
[GitHub] [incubator-mxnet] zachgk closed issue #15143: dmlc::type_name_helper specialization of mxnet::tuple should not be disabled for MSVC
zachgk closed issue #15143: dmlc::type_name_helper specialization of mxnet::tuple should not be disabled for MSVC URL: https://github.com/apache/incubator-mxnet/issues/15143 This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-mxnet] zachgk closed issue #14116: Failure in generated op.h in version 1.3.1
zachgk closed issue #14116: Failure in generated op.h in version 1.3.1 URL: https://github.com/apache/incubator-mxnet/issues/14116 This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-mxnet] zachgk merged pull request #15144: Relax Visual Studio version constraint in the specialization of `dmlc::type_name_helper` for `DT=mxnet::Tuple`
zachgk merged pull request #15144: Relax Visual Studio version constraint in the specialization of `dmlc::type_name_helper` for `DT=mxnet::Tuple` URL: https://github.com/apache/incubator-mxnet/pull/15144 This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-mxnet] zachgk commented on issue #15797: how can I concat two dataset with a sepcial ratio?
zachgk commented on issue #15797: how can I concat two dataset with a sepcial ratio? URL: https://github.com/apache/incubator-mxnet/issues/15797#issuecomment-533698381 @crossLi Were you able to get the custom data loader working? This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-mxnet] Jerryzcn commented on a change in pull request #16215: New ops for RCNN + old ops improvements for RCNN
Jerryzcn commented on a change in pull request #16215: New ops for RCNN + old
ops improvements for RCNN
URL: https://github.com/apache/incubator-mxnet/pull/16215#discussion_r326790569
##
File path: src/operator/contrib/bounding_box-inl.h
##
@@ -787,6 +787,284 @@ void BipartiteMatchingBackward(const nnvm::NodeAttrs&
attrs,
});
}
+
+inline bool BoxEncodeShape(const nnvm::NodeAttrs& attrs,
+ mxnet::ShapeVector *in_attrs,
+ mxnet::ShapeVector *out_attrs) {
+ CHECK_EQ(in_attrs->size(), 6U);
+ CHECK_EQ(out_attrs->size(), 2U);
+ mxnet::TShape& sshape = (*in_attrs)[0];
+ mxnet::TShape& mshape = (*in_attrs)[1];
+ mxnet::TShape& ashape = (*in_attrs)[2];
+ mxnet::TShape& rshape = (*in_attrs)[3];
+
+ CHECK_EQ(sshape.ndim(), 2)
+<< "samples shape must have dim == 2, "
+<< sshape.ndim() << " provided";
+
+ CHECK_GE(mshape.ndim(), 2)
+<< "matches shape must have dim == 2, "
+<< mshape.ndim() << " provided";
+
+ CHECK_GE(ashape.ndim(), 3)
+<< "matches shape must have dim == 3, "
+<< ashape.ndim() << " provided";
+ int ldim = ashape[ashape.ndim() - 1];
+ CHECK_EQ(ldim, 4)
+<< "last dimension of anchors must be 4, "
+<< ldim << " provided";
+
+ CHECK_GE(rshape.ndim(), 3)
+<< "refs shape must have dim == 3, "
+<< ashape.ndim() << " provided";
+ ldim = rshape[rshape.ndim() - 1];
+ CHECK_EQ(ldim, 4)
+<< "last dimension of anchors must be 4, "
+<< ldim << " provided";
+
+ // asign input shape
+ SHAPE_ASSIGN_CHECK(*in_attrs, 4, mshadow::Shape1(4));
+ SHAPE_ASSIGN_CHECK(*in_attrs, 5, mshadow::Shape1(4));
+
+ // assign output shape
+ mxnet::TShape oshape = ashape;
+ SHAPE_ASSIGN_CHECK(*out_attrs, 0, oshape);
+ SHAPE_ASSIGN_CHECK(*out_attrs, 1, oshape);
+ return shape_is_known(oshape);
+}
+
+struct box_encode {
+ template
+ MSHADOW_XINLINE static void Map(int i, DType *out_targets, DType *out_masks,
Review comment:
understood.
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org
With regards,
Apache Git Services
[GitHub] [incubator-mxnet] Jerryzcn commented on a change in pull request #16215: New ops for RCNN + old ops improvements for RCNN
Jerryzcn commented on a change in pull request #16215: New ops for RCNN + old
ops improvements for RCNN
URL: https://github.com/apache/incubator-mxnet/pull/16215#discussion_r326790631
##
File path: src/operator/contrib/bounding_box-inl.h
##
@@ -787,6 +787,284 @@ void BipartiteMatchingBackward(const nnvm::NodeAttrs&
attrs,
});
}
+
+inline bool BoxEncodeShape(const nnvm::NodeAttrs& attrs,
+ mxnet::ShapeVector *in_attrs,
+ mxnet::ShapeVector *out_attrs) {
+ CHECK_EQ(in_attrs->size(), 6U);
+ CHECK_EQ(out_attrs->size(), 2U);
+ mxnet::TShape& sshape = (*in_attrs)[0];
+ mxnet::TShape& mshape = (*in_attrs)[1];
+ mxnet::TShape& ashape = (*in_attrs)[2];
+ mxnet::TShape& rshape = (*in_attrs)[3];
+
+ CHECK_EQ(sshape.ndim(), 2)
+<< "samples shape must have dim == 2, "
+<< sshape.ndim() << " provided";
+
+ CHECK_GE(mshape.ndim(), 2)
+<< "matches shape must have dim == 2, "
+<< mshape.ndim() << " provided";
+
+ CHECK_GE(ashape.ndim(), 3)
+<< "matches shape must have dim == 3, "
+<< ashape.ndim() << " provided";
+ int ldim = ashape[ashape.ndim() - 1];
+ CHECK_EQ(ldim, 4)
+<< "last dimension of anchors must be 4, "
+<< ldim << " provided";
+
+ CHECK_GE(rshape.ndim(), 3)
+<< "refs shape must have dim == 3, "
+<< ashape.ndim() << " provided";
+ ldim = rshape[rshape.ndim() - 1];
+ CHECK_EQ(ldim, 4)
+<< "last dimension of anchors must be 4, "
+<< ldim << " provided";
+
+ // asign input shape
+ SHAPE_ASSIGN_CHECK(*in_attrs, 4, mshadow::Shape1(4));
+ SHAPE_ASSIGN_CHECK(*in_attrs, 5, mshadow::Shape1(4));
+
+ // assign output shape
+ mxnet::TShape oshape = ashape;
+ SHAPE_ASSIGN_CHECK(*out_attrs, 0, oshape);
+ SHAPE_ASSIGN_CHECK(*out_attrs, 1, oshape);
+ return shape_is_known(oshape);
+}
+
+struct box_encode {
+ template
+ MSHADOW_XINLINE static void Map(int i, DType *out_targets, DType *out_masks,
+ const DType *samples, const DType *matches,
+ const DType *anchors, const DType *refs,
+ const DType *means, const DType *stds,
+ const int m, const int n) {
+int j = i / n;
+int match = matches[i];
+// xmin: 0, ymin:1, xmax: 2, ymax: 3
+// x:0, y:1, w:2, h:3
+int ref_index = (j * m + match) * 4;
+DType ref_xmin = refs[ref_index + 0];
+DType ref_ymin = refs[ref_index + 1];
+DType ref_width = refs[ref_index + 2] - ref_xmin;
+DType ref_height = refs[ref_index + 3] - ref_ymin;
+DType ref_x = ref_xmin + ref_width * 0.5;
+DType ref_y = ref_ymin + ref_height * 0.5;
+int a_index = i * 4;
+DType a_xmin = anchors[a_index + 0];
+DType a_ymin = anchors[a_index + 1];
+DType a_width = anchors[a_index + 2] - a_xmin;
+DType a_height = anchors[a_index + 3] - a_ymin;
+DType a_x = a_xmin + a_width * 0.5;
+DType a_y = a_ymin + a_height * 0.5;
+DType valid = samples[i] > 0.5 ? 1.0 : 0.0;
+out_masks[a_index + 0] = valid;
+out_masks[a_index + 1] = valid;
+out_masks[a_index + 2] = valid;
+out_masks[a_index + 3] = valid;
+out_targets[a_index + 0] = valid > static_cast(0.5) ?
+((ref_x - a_x) / a_width - static_cast(means[0])) /
+static_cast(stds[0]) : static_cast(0.0);
+out_targets[a_index + 1] = valid > static_cast(0.5) ?
+((ref_y - a_y) / a_height - static_cast(means[1])) /
+static_cast(stds[1]) : static_cast(0.0);
+out_targets[a_index + 2] = valid > static_cast(0.5) ?
+(log(ref_width / a_width) - static_cast(means[2])) /
+static_cast(stds[2]) : static_cast(0.0);
+out_targets[a_index + 3] = valid > static_cast(0.5) ?
+(log(ref_height / a_height) - static_cast(means[3])) /
+static_cast(stds[3]) : static_cast(0.0);
+ }
+};
+
+template
+void BoxEncodeForward(const nnvm::NodeAttrs& attrs,
+ const OpContext& ctx,
+ const std::vector& inputs,
+ const std::vector& req,
+ const std::vector& outputs) {
+ using namespace mshadow;
+ using namespace mshadow::expr;
+ using namespace mxnet_op;
+ CHECK_EQ(inputs.size(), 6U);
+ CHECK_EQ(outputs.size(), 2U);
+ Stream *s = ctx.get_stream();
+ // samples, matches, anchors, refs, means, stds
+ mxnet::TShape anchor_shape = inputs[2].shape_;
+ int loop_size = anchor_shape.ProdShape(0, 2);
+ int b = anchor_shape[0];
+ int n = anchor_shape[1];
+ int m = inputs[3].shape_[1];
+ MSHADOW_REAL_TYPE_SWITCH(outputs[0].type_flag_, DType, {
+Tensor samples = inputs[0]
+ .get_with_shape(Shape2(b, n), s);
+Tensor matches = inputs[1]
+ .get_with_shape(Shape2(b, n), s);
+Tensor anchors = inputs[2]
+ .get_with_shape(Shape3(b, n, 4), s);
+Tensor refs = inputs[3]
+ .get_with_shape(Shape3(b, m, 4), s);
+Tensor means = inputs[4]
+ .get_with_shape(Shape1(4), s);
[incubator-mxnet-site] branch asf-site updated: Bump the publish timestamp.
This is an automated email from the ASF dual-hosted git repository. aaronmarkham pushed a commit to branch asf-site in repository https://gitbox.apache.org/repos/asf/incubator-mxnet-site.git The following commit(s) were added to refs/heads/asf-site by this push: new 2c2fb24 Bump the publish timestamp. 2c2fb24 is described below commit 2c2fb2401c0b0b7e85e1a90d807f7cf1784f91cb Author: mxnet-ci AuthorDate: Fri Sep 20 18:41:41 2019 + Bump the publish timestamp. --- date.txt | 1 + 1 file changed, 1 insertion(+) diff --git a/date.txt b/date.txt new file mode 100644 index 000..98174a3 --- /dev/null +++ b/date.txt @@ -0,0 +1 @@ +Fri Sep 20 18:41:41 UTC 2019
[GitHub] [incubator-mxnet] zachgk commented on issue #16224: mxnet.base.MXNetError: [13:43:35] C:\Jenkins\workspace\mxnet-tag\mxnet\src\ndarray\ndarray.cc:1755: Check failed: fi->Read(data) Invalid ND
zachgk commented on issue #16224: mxnet.base.MXNetError: [13:43:35] C:\Jenkins\workspace\mxnet-tag\mxnet\src\ndarray\ndarray.cc:1755: Check failed: fi->Read(data) Invalid NDArray file format while loading resnet-18 URL: https://github.com/apache/incubator-mxnet/issues/16224#issuecomment-533666873 I ran this and it seemed to work fine. The mx.test_utils.download directory tries to download the files to the local directory. Can you try navigating to an empty directory before running this? Can you also share your environment info: ``` What to do: 1. Download the diagnosis script from https://raw.githubusercontent.com/apache/incubator-mxnet/master/tools/diagnose.py 2. Run the script using `python diagnose.py` and paste its output here. ``` This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-mxnet] zachgk commented on issue #16227: How to use a pre-trained model in model zoo as a sub-block in my own model?
zachgk commented on issue #16227: How to use a pre-trained model in model zoo as a sub-block in my own model? URL: https://github.com/apache/incubator-mxnet/issues/16227#issuecomment-533663914 You should use Mymodel.name_scope() when defining the child blocks. Take a look at the documentation in http://beta.mxnet.io/api/gluon/mxnet.gluon.nn.Block.html. Also, use model.save_parameters as save_params is deprecated. This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-mxnet] zachgk commented on issue #9656: Sparse Tensor support in scala api
zachgk commented on issue #9656: Sparse Tensor support in scala api URL: https://github.com/apache/incubator-mxnet/issues/9656#issuecomment-533658182 Sparse Support was added in https://github.com/apache/incubator-mxnet/pull/15378 This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-mxnet] zachgk closed issue #9656: Sparse Tensor support in scala api
zachgk closed issue #9656: Sparse Tensor support in scala api URL: https://github.com/apache/incubator-mxnet/issues/9656 This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-mxnet] zachgk commented on issue #15144: Relax Visual Studio version constraint in the specialization of `dmlc::type_name_helper` for `DT=mxnet::Tuple`
zachgk commented on issue #15144: Relax Visual Studio version constraint in the specialization of `dmlc::type_name_helper` for `DT=mxnet::Tuple` URL: https://github.com/apache/incubator-mxnet/pull/15144#issuecomment-533651460 @QueensGambit Thanks for reviewing. You can actually use the official approve GitHub approve button under Files Changed -> Review Changes -> Approve even if you are not a committer. @anirudh2290 @junrushao1994 @eric-haibin-lin Can any of you help review this PR? This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-mxnet] zachgk commented on issue #16144: added -DCMAKE_BUILD_TYPE=Release to docs for building from source
zachgk commented on issue #16144: added -DCMAKE_BUILD_TYPE=Release to docs for building from source URL: https://github.com/apache/incubator-mxnet/pull/16144#issuecomment-533647577 @QueensGambit There are merge conflicts since we released our new website (https://github.com/apache/incubator-mxnet/pull/15883). Can you rebase? This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-mxnet] drivanov commented on a change in pull request #16218: Improving performance of argmax operator
drivanov commented on a change in pull request #16218: Improving performance of
argmax operator
URL: https://github.com/apache/incubator-mxnet/pull/16218#discussion_r326727093
##
File path: src/operator/tensor/broadcast_reduce_op.h
##
@@ -556,6 +556,162 @@ inline bool ReduceAxesOpForwardStorage(const
nnvm::NodeAttrs& attrs,
return dispatched;
}
+struct argmax {
+ template
+ MSHADOW_XINLINE static void Map(int i, const int nWorkers, const DType
*in_data, DType *out_data,
Review comment:
OK, I will do that.
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org
With regards,
Apache Git Services
[GitHub] [incubator-mxnet] drivanov commented on a change in pull request #16218: Improving performance of argmax operator
drivanov commented on a change in pull request #16218: Improving performance of
argmax operator
URL: https://github.com/apache/incubator-mxnet/pull/16218#discussion_r326727093
##
File path: src/operator/tensor/broadcast_reduce_op.h
##
@@ -556,6 +556,162 @@ inline bool ReduceAxesOpForwardStorage(const
nnvm::NodeAttrs& attrs,
return dispatched;
}
+struct argmax {
+ template
+ MSHADOW_XINLINE static void Map(int i, const int nWorkers, const DType
*in_data, DType *out_data,
Review comment:
OK, I will do that.
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org
With regards,
Apache Git Services
[GitHub] [incubator-mxnet] drivanov edited a comment on issue #16178: [WIP]improving argmax perf
drivanov edited a comment on issue #16178: [WIP]improving argmax perf URL: https://github.com/apache/incubator-mxnet/pull/16178#issuecomment-533637642 @access2rohit: It seems that to pass following test from /work/mxnet/julia/test/unittest/ndarray.jl, lines1511-1531: ``` function test_argmax() . . . @info "NDArray::argmax" @info "NDArray::argmax::NaN" let A = [1. 5 3; NaN 2 6] x = NDArray(A) @test copy(argmax(x, dims = 1)) == [1 1 2] @test copy(argmax(x, dims = 2)) == reshape([2, 3], :, 1) end end ``` We have to skip `NaN`'s. As far as I understand, in that case the outputs of `nd.argmax` and `np.nanarmax` should be identical and we will non need `nd.nanargmax`. Is it correct? BTW, what would be correct output of `nd.argmax` when all elements of vector will be `NaN`'s? ``` A = [1. 5 3; NaN NaN NaN] ``` This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-mxnet] drivanov commented on issue #16178: [WIP]improving argmax perf
drivanov commented on issue #16178: [WIP]improving argmax perf URL: https://github.com/apache/incubator-mxnet/pull/16178#issuecomment-533637642 @access2rohit: It seems that to pass following test fron /work/mxnet/julia/test/unittest/ndarray.jl, lines1511-1531: ``` function test_argmax() . . . @info "NDArray::argmax" @info "NDArray::argmax::NaN" let A = [1. 5 3; NaN 2 6] x = NDArray(A) @test copy(argmax(x, dims = 1)) == [1 1 2] @test copy(argmax(x, dims = 2)) == reshape([2, 3], :, 1) end end ``` We have to skip `NaN`s. As far as I understand, in that case the outputs of `nd.argmax` and `np.nanarmax` should be identical and we will non need `nd.nanargmax`. Is it correct? This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-mxnet] kshitij12345 commented on issue #15331: [fix] missing input log higher order.
kshitij12345 commented on issue #15331: [fix] missing input log higher order. URL: https://github.com/apache/incubator-mxnet/pull/15331#issuecomment-533627735 @sxjscience I am not sure about how to test this. I was expecting that this missing input will cause problem with computing of higher order. Tried for 3rd order, which was computed successfully. For fourth order, `Operator _backward_mul is non-differentiable because it didn't register FGradient attribute` (different problem). So I am not very sure as to how to write a test case. This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-mxnet] Caenorst commented on issue #16122: Add fast implementation of LARS
Caenorst commented on issue #16122: Add fast implementation of LARS URL: https://github.com/apache/incubator-mxnet/pull/16122#issuecomment-533627265 I don't understand the errors in the CI, can somebody help me ? This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-mxnet] Caenorst commented on a change in pull request #16122: Add fast implementation of LARS
Caenorst commented on a change in pull request #16122: Add fast implementation
of LARS
URL: https://github.com/apache/incubator-mxnet/pull/16122#discussion_r326710971
##
File path: tests/python/gpu/test_optimizer.py
##
@@ -0,0 +1,93 @@
+# Licensed to the Apache Software Foundation (ASF) under one
+# or more contributor license agreements. See the NOTICE file
+# distributed with this work for additional information
+# regarding copyright ownership. The ASF licenses this file
+# to you under the Apache License, Version 2.0 (the
+# "License"); you may not use this file except in compliance
+# with the License. You may obtain a copy of the License at
+#
+# http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing,
+# software distributed under the License is distributed on an
+# "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
+# KIND, either express or implied. See the License for the
+# specific language governing permissions and limitations
+# under the License.
+
+import mxnet as mx
+
+import sys
+import os
+curr_path = os.path.dirname(os.path.abspath(os.path.expanduser(__file__)))
+sys.path.insert(0, os.path.join(curr_path, '../unittest'))
+from common import setup_module, with_seed
+
+# This script is testing the efficiency of LARS
+# We are training LeNet-5 at batch-size 8000 in 10 epochs above 98% accuracy
+# Which is not doable with simple SGD + momentum (from what have been tested
so far)
+
+def lenet5():
+"""LeNet-5 Symbol"""
+#pylint: disable=no-member
+data = mx.sym.Variable('data')
+conv1 = mx.sym.Convolution(data=data, kernel=(5, 5), num_filter=20)
+tanh1 = mx.sym.Activation(data=conv1, act_type="tanh")
+pool1 = mx.sym.Pooling(data=tanh1, pool_type="max",
+ kernel=(2, 2), stride=(2, 2))
+# second conv
+conv2 = mx.sym.Convolution(data=pool1, kernel=(5, 5), num_filter=50)
+tanh2 = mx.sym.Activation(data=conv2, act_type="tanh")
+pool2 = mx.sym.Pooling(data=tanh2, pool_type="max",
+ kernel=(2, 2), stride=(2, 2))
+# first fullc
+flatten = mx.sym.Flatten(data=pool2)
+fc1 = mx.sym.FullyConnected(data=flatten, num_hidden=500)
+tanh3 = mx.sym.Activation(data=fc1, act_type="tanh")
+# second fullc
+fc2 = mx.sym.FullyConnected(data=tanh3, num_hidden=10)
+# loss
+lenet = mx.sym.SoftmaxOutput(data=fc2, name='softmax')
+#pylint: enable=no-member
+return lenet
+
+@with_seed()
+def test_lars():
Review comment:
So, I'm already testing the MXNet Ops in
https://github.com/apache/incubator-mxnet/pull/16122/files#diff-4758fb9329d438de2836db2634a8f5f7R270-R422.
which are the only non-python part of the optimizer, what my test is adding is
to show that it's properly behaving: i.e allow you to train a network with a
bigger batch. Making a fully python of it wouldn't help me testing that (the
python could be wrong as well, let's say if I misunderstood the publication)
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org
With regards,
Apache Git Services
[GitHub] [incubator-mxnet] marcoabreu commented on a change in pull request #15855: [Numpy] cumprod
marcoabreu commented on a change in pull request #15855: [Numpy] cumprod
URL: https://github.com/apache/incubator-mxnet/pull/15855#discussion_r326710965
##
File path: ci/docker/runtime_functions.sh
##
@@ -367,6 +367,7 @@ build_ubuntu_cpu_openblas() {
set -ex
export CC="gcc"
export CXX="g++"
+pip3 install --user psutil
Review comment:
How about you add Ubuntu tvm to the CPU dockerfile? I mean if we want to run
tvm cpu tests, we should also install it, right?
No, please don't separate it.
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org
With regards,
Apache Git Services
[GitHub] [incubator-mxnet] hzfan commented on a change in pull request #15855: [Numpy] cumprod
hzfan commented on a change in pull request #15855: [Numpy] cumprod
URL: https://github.com/apache/incubator-mxnet/pull/15855#discussion_r326710119
##
File path: ci/docker/runtime_functions.sh
##
@@ -367,6 +367,7 @@ build_ubuntu_cpu_openblas() {
set -ex
export CC="gcc"
export CXX="g++"
+pip3 install --user psutil
Review comment:
Thank you for the suggestion. I tried it and found that
`Dockerfile.build.ubuntu_gpu_cu*` runs `ubuntu_tvm.sh`, but
`Dockerfile.build.ubuntu_cpu` does not run it. So I think doing it in
`ubuntu_tvm.sh` may not fix the error in unix-cpu.
Can I separate `ubuntu_python.sh` into two files, one for arm device and the
other for common x86-64 device? This worked for me.
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org
With regards,
Apache Git Services
[GitHub] [incubator-mxnet] drivanov commented on issue #16218: Improving performance of argmax operator
drivanov commented on issue #16218: Improving performance of argmax operator URL: https://github.com/apache/incubator-mxnet/pull/16218#issuecomment-533624848 > It seems this PR is working on the same operator as #16178. Can you run a profiling using https://github.com/apache/incubator-mxnet/tree/master/benchmark/opperf and show the performance gain of your change? I will do it and will collect performance data for all three version: existing one, #16178 and #16218 This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-mxnet] drivanov commented on a change in pull request #16218: Improving performance of argmax operator
drivanov commented on a change in pull request #16218: Improving performance of
argmax operator
URL: https://github.com/apache/incubator-mxnet/pull/16218#discussion_r326707791
##
File path: src/operator/tensor/broadcast_reduce_op.h
##
@@ -556,6 +556,162 @@ inline bool ReduceAxesOpForwardStorage(const
nnvm::NodeAttrs& attrs,
return dispatched;
}
+struct argmax {
+ template
+ MSHADOW_XINLINE static void Map(int i, const int nWorkers, const DType
*in_data, DType *out_data,
+ size_t nSteps, size_t step, size_t shift, void *pIdxStorage,
bool use_uint16) {
+// i - index of launched thread
+// nWorkers - number of threads, assigned to work on one row/column
+// iw - index of current thread among workers assigned to the same vector
+int iw = 0;
+const DType *pCurr = in_data;
+if (nWorkers > 1) {
+ // in - the vector number which current thread is assigned to
+ const auto in = i / nWorkers;
+ iw = i % nWorkers;
+ pCurr += in % step + shift * (in / step) + iw * step;
+ nSteps = (nSteps + nWorkers - 1 - iw) / nWorkers;
+ step *= nWorkers;
+} else {
+ pCurr += i % step + shift * (i / step);
+}
+
+int maxIdx = 0;
+DType maxVal = *pCurr;
+for (size_t j = 1; j < nSteps; ++j) {
+ if (maxVal < *(pCurr += step)) {
+maxVal = *pCurr;
+maxIdx = j;
+ }
+}
+
+if (nWorkers > 1) {
+ // saving index of best element found by current thread
+ if (use_uint16) {
+*(static_cast(pIdxStorage) + i) = maxIdx * nWorkers + iw;
+ } else {
+*(static_cast(pIdxStorage) + i) = maxIdx * nWorkers + iw;
+ }
+} else {
+ out_data[i] = maxIdx;// output of argmax
+}
+ }
+};
+
+struct argmax_reduce {
+ template
+ MSHADOW_XINLINE static void Map(int i, const int nWorkers, const DType
*in_data, DType *out_data,
+ const size_t step, const size_t shift, void *pIdx, const bool
use_uin16) {
+const DType *pCurr = in_data + i % step + shift * (i / step);
+int maxIdx;
+if (use_uin16) {
+ const auto *pIdxStorage = static_cast(pIdx);
+ maxIdx = *(pIdxStorage += i * nWorkers);
+ DType maxVal = *(pCurr + step * maxIdx);
+ for (int j = 1; j < nWorkers; ++j) {
+const auto val = *(pCurr + step * pIdxStorage[j]);
+if (maxVal < val) {
+ maxVal = val;
+ maxIdx = pIdxStorage[j];
+}
+ }
+} else {
+ const auto *pIdxStorage = static_cast(pIdx);
+ maxIdx = *(pIdxStorage += i * nWorkers);
+ DType maxVal = *(pCurr + step * maxIdx);
+ for (int j = 1; j < nWorkers; ++j) {
+const auto val = *(pCurr + step * pIdxStorage[j]);
+if (maxVal < val) {
+ maxVal = val;
+ maxIdx = pIdxStorage[j];
+}
+ }
+}
+
+out_data[i] = maxIdx;
+ }
+};
+
+template
+DType *AllocateDTypeMemory(const OpContext& ctx, const size_t num_items) {
+ const size_t memSize = num_items * sizeof(DType);
+ mshadow::Tensor workspace =
+ctx.requested[0].get_space_typed(
+ mshadow::Shape1(memSize), ctx.get_stream());
+ return reinterpret_cast(workspace.dptr_);
+}
+
+template
+void ArgMax(const OpContext& ctx, const TShape &shape, int axis, size_t step,
size_t shift,
+const int nWorkers, void *pIdxStorage, bool use_uin16,
+const TBlob& input, const TBlob& output) {
+ using namespace mxnet_op;
+ const auto pIn = input.dptr();
+ auto *pOut = output.dptr();
+ const auto nSize = shape[axis];
+ const auto num_threads = shape.Size() / nSize;
+ mshadow::Stream *s = ctx.get_stream();
+ Kernel::Launch(s, num_threads * nWorkers, nWorkers, pIn, pOut,
nSize,
+ step, shift, pIdxStorage, use_uin16);
+ if (nWorkers > 1) {
+Kernel::Launch(s, num_threads, nWorkers, pIn, pOut,
+ step, shift, pIdxStorage, use_uin16);
+ }
+}
+
+template
+int DefineNumbWorkers(const TShape &shape, int axis);
+
+template
+void ArgMax(const nnvm::NodeAttrs& attrs,
+const OpContext& ctx,
+const std::vector& inputs,
+const std::vector& req,
+const std::vector& outputs) {
+ const ReduceAxisParam& param = nnvm::get(attrs.parsed);
+ if (!param.axis) LOG(FATAL) << "Global reduction not supported yet";
+
+ auto shape = inputs[0].shape_;
+ auto axis = CheckAxis(param.axis.value(), shape.ndim());
+ if (shape.ndim() == 1)
+shape = AxisShapeCompact(shape, &axis, true);
+
+ void *pIdxMemory = nullptr;
+ auto nWorkers = DefineNumbWorkers(shape, axis);
+ bool use_uint16 = false;
+ while (nWorkers > 1) {
+const size_t num_items = nWorkers * shape.Size() / shape[axis];
+pIdxMemory = AllocateDTypeMemory(ctx, num_items);
+if (pIdxMemory)
+ break;
+
+// Check if indexes can be stored in uint16 format
+if (shape[axis] <= UINT16_MAX) {
+ pIdxMemory = AllocateDTypeMemory(ctx, num_items);
+
[GitHub] [incubator-mxnet] drivanov commented on a change in pull request #16218: Improving performance of argmax operator
drivanov commented on a change in pull request #16218: Improving performance of
argmax operator
URL: https://github.com/apache/incubator-mxnet/pull/16218#discussion_r326707461
##
File path: src/operator/tensor/broadcast_reduce_op.h
##
@@ -556,6 +556,162 @@ inline bool ReduceAxesOpForwardStorage(const
nnvm::NodeAttrs& attrs,
return dispatched;
}
+struct argmax {
+ template
+ MSHADOW_XINLINE static void Map(int i, const int nWorkers, const DType
*in_data, DType *out_data,
Review comment:
OK, I will do that.
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org
With regards,
Apache Git Services
[GitHub] [incubator-mxnet] drivanov commented on a change in pull request #16218: Improving performance of argmax operator
drivanov commented on a change in pull request #16218: Improving performance of
argmax operator
URL: https://github.com/apache/incubator-mxnet/pull/16218#discussion_r326704192
##
File path: src/operator/tensor/broadcast_reduce_op.h
##
@@ -556,6 +556,162 @@ inline bool ReduceAxesOpForwardStorage(const
nnvm::NodeAttrs& attrs,
return dispatched;
}
+struct argmax {
+ template
+ MSHADOW_XINLINE static void Map(int i, const int nWorkers, const DType
*in_data, DType *out_data,
+ size_t nSteps, size_t step, size_t shift, void *pIdxStorage,
bool use_uint16) {
+// i - index of launched thread
+// nWorkers - number of threads, assigned to work on one row/column
+// iw - index of current thread among workers assigned to the same vector
+int iw = 0;
+const DType *pCurr = in_data;
+if (nWorkers > 1) {
+ // in - the vector number which current thread is assigned to
+ const auto in = i / nWorkers;
+ iw = i % nWorkers;
+ pCurr += in % step + shift * (in / step) + iw * step;
+ nSteps = (nSteps + nWorkers - 1 - iw) / nWorkers;
+ step *= nWorkers;
+} else {
+ pCurr += i % step + shift * (i / step);
+}
+
+int maxIdx = 0;
+DType maxVal = *pCurr;
+for (size_t j = 1; j < nSteps; ++j) {
+ if (maxVal < *(pCurr += step)) {
+maxVal = *pCurr;
+maxIdx = j;
+ }
+}
+
+if (nWorkers > 1) {
+ // saving index of best element found by current thread
+ if (use_uint16) {
+*(static_cast(pIdxStorage) + i) = maxIdx * nWorkers + iw;
+ } else {
+*(static_cast(pIdxStorage) + i) = maxIdx * nWorkers + iw;
+ }
+} else {
+ out_data[i] = maxIdx;// output of argmax
+}
+ }
+};
+
+struct argmax_reduce {
+ template
+ MSHADOW_XINLINE static void Map(int i, const int nWorkers, const DType
*in_data, DType *out_data,
+ const size_t step, const size_t shift, void *pIdx, const bool
use_uin16) {
+const DType *pCurr = in_data + i % step + shift * (i / step);
+int maxIdx;
+if (use_uin16) {
Review comment:
Yes, I also thought about it, and I even wrote one. Unfortunately, I could
not compile this template. Perhaps, I will try one more time.
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org
With regards,
Apache Git Services
[GitHub] [incubator-mxnet] drivanov commented on a change in pull request #16218: Improving performance of argmax operator
drivanov commented on a change in pull request #16218: Improving performance of
argmax operator
URL: https://github.com/apache/incubator-mxnet/pull/16218#discussion_r326699897
##
File path: src/operator/tensor/broadcast_reduce_op_index.cu
##
@@ -43,5 +43,18 @@ NNVM_REGISTER_OP(pick)
NNVM_REGISTER_OP(_backward_pick)
.set_attr("FCompute", PickOpBackward);
+template<>
+int DefineNumbWorkers(const TShape &shape, int axis) {
+ const auto nSteps = static_cast(shape[axis]);
+ const auto nThreads = shape.Size()/nSteps;
+ if (nThreads > nSteps)
+return 1;
+
+ const auto a = static_cast(nSteps)/nThreads;
+ const auto b = log2f(a);
+ const auto numbWorkers = pow(2, (b * 5 + 28)/11);
Review comment:
This is just a heuristic. Experimenting with different values `(shape, axis,
numbWorkers)`, I accumulated a lot of data, it turned out that they almost
perfectly correspond to this formula.
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org
With regards,
Apache Git Services
[GitHub] [incubator-mxnet] drivanov commented on a change in pull request #16218: Improving performance of argmax operator
drivanov commented on a change in pull request #16218: Improving performance of
argmax operator
URL: https://github.com/apache/incubator-mxnet/pull/16218#discussion_r326699897
##
File path: src/operator/tensor/broadcast_reduce_op_index.cu
##
@@ -43,5 +43,18 @@ NNVM_REGISTER_OP(pick)
NNVM_REGISTER_OP(_backward_pick)
.set_attr("FCompute", PickOpBackward);
+template<>
+int DefineNumbWorkers(const TShape &shape, int axis) {
+ const auto nSteps = static_cast(shape[axis]);
+ const auto nThreads = shape.Size()/nSteps;
+ if (nThreads > nSteps)
+return 1;
+
+ const auto a = static_cast(nSteps)/nThreads;
+ const auto b = log2f(a);
+ const auto numbWorkers = pow(2, (b * 5 + 28)/11);
Review comment:
This is just a heuristic. Experimenting with different values `(shape, axis,
numbWorkers)`, I accumulated a lot of data, it turned out that they almost
perfectly corresponded to this formula.
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org
With regards,
Apache Git Services
[GitHub] [incubator-mxnet] leezu commented on a change in pull request #16207: Bump numpy version >=1.17.0
leezu commented on a change in pull request #16207: Bump numpy version >=1.17.0
URL: https://github.com/apache/incubator-mxnet/pull/16207#discussion_r326661047
##
File path: python/setup.py
##
@@ -30,7 +30,11 @@
else:
from setuptools import setup
from setuptools.extension import Extension
-kwargs = {'install_requires': ['numpy>=1.17.0,<2.0.0',
'requests>=2.20.0,<3', 'graphviz<0.9.0,>=0.8.1'], 'zip_safe': False}
+kwargs = {
+'install_requires': ['requests>=2.20.0,<3', 'graphviz<0.9.0,>=0.8.1']
+.append('numpy>=1.17.0,<2.0.0' if sys.version_info[0] > 2 else
'numpy>1.16.0,<2.0.0'),
Review comment:
Why not use https://www.python.org/dev/peps/pep-0508/#environment-markers ?
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org
With regards,
Apache Git Services
[GitHub] [incubator-mxnet] TaoLv commented on issue #16143: Failure of MKL-DNN Convolution from C API
TaoLv commented on issue #16143: Failure of MKL-DNN Convolution from C API URL: https://github.com/apache/incubator-mxnet/issues/16143#issuecomment-533572996 Sorry for the delay @matteosal . I got trapped by other stuff this week. Will look into the python script and get back to you next week. Thanks for your patience. This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[incubator-mxnet] branch mkldnn-v1.0 updated: [mkldnn-v1.0] Add MKL-DNN activation (#16195)
This is an automated email from the ASF dual-hosted git repository.
taolv pushed a commit to branch mkldnn-v1.0
in repository https://gitbox.apache.org/repos/asf/incubator-mxnet.git
The following commit(s) were added to refs/heads/mkldnn-v1.0 by this push:
new f930baa [mkldnn-v1.0] Add MKL-DNN activation (#16195)
f930baa is described below
commit f930baa533d24497188e33c163dbb1f36707c336
Author: rongzha1
AuthorDate: Fri Sep 20 22:09:19 2019 +0800
[mkldnn-v1.0] Add MKL-DNN activation (#16195)
* add mkldnn act; pass lint; pass mnist training
* make bwd as private member
---
src/operator/nn/activation.cc | 18 ++---
src/operator/nn/mkldnn/mkldnn_act-inl.h | 40 +++---
src/operator/nn/mkldnn/mkldnn_act.cc| 130
src/operator/nn/mkldnn/mkldnn_ops-inl.h | 15 ++--
4 files changed, 75 insertions(+), 128 deletions(-)
diff --git a/src/operator/nn/activation.cc b/src/operator/nn/activation.cc
index 5abb667..f238e8f 100644
--- a/src/operator/nn/activation.cc
+++ b/src/operator/nn/activation.cc
@@ -27,10 +27,10 @@
#include "./activation-inl.h"
#include "../mshadow_op.h"
#include "../tensor/elemwise_unary_op.h"
-#if MXNET_USE_MKLDNN == 1
+#if MXNET_USE_MKLDNN == 100
#include "./mkldnn/mkldnn_base-inl.h"
#include "./mkldnn/mkldnn_ops-inl.h"
-#endif // MXNET_USE_MKLDNN == 1
+#endif // MXNET_USE_MKLDNN == 100
#include "../operator_common.h"
#include "../../common/utils.h"
@@ -91,7 +91,7 @@ struct ActivationGrad {
}
};
-#if MXNET_USE_MKLDNN == 1
+#if MXNET_USE_MKLDNN == 100
static void ActivationComputeExCPU(const nnvm::NodeAttrs& attrs,
const OpContext& ctx,
const std::vector& inputs,
@@ -150,7 +150,7 @@ inline static bool BackwardActStorageType(const
nnvm::NodeAttrs& attrs,
return MKLDNNStorageType(attrs, dev_mask, SupportMKLDNNAct(param),
dispatch_mode, in_attrs, out_attrs);
}
-#endif // MXNET_USE_MKLDNN == 1
+#endif // MXNET_USE_MKLDNN == 100
MXNET_OPERATOR_REGISTER_UNARY(Activation)
@@ -167,7 +167,7 @@ The following activation functions are supported:
)code" ADD_FILELINE)
.set_attr_parser(ParamParser)
-#if MXNET_USE_MKLDNN == 1
+#if MXNET_USE_MKLDNN == 100
.set_attr("FInferStorageType", ActivationStorageType)
#endif
.set_attr("FListOutputNames",
@@ -175,7 +175,7 @@ The following activation functions are supported:
return std::vector{"output"};
})
.set_attr("FCompute", ActivationCompute)
-#if MXNET_USE_MKLDNN == 1
+#if MXNET_USE_MKLDNN == 100
.set_attr("TIsMKLDNN", true)
.set_attr("FComputeEx", ActivationComputeExCPU)
#endif
@@ -189,7 +189,7 @@ NNVM_REGISTER_OP(_backward_Activation)
})
.set_num_outputs(1)
.set_attr("TIsBackward", true)
-#if MXNET_USE_MKLDNN == 1
+#if MXNET_USE_MKLDNN == 100
.set_attr("FInferStorageType", BackwardActStorageType)
#endif
.set_attr("FInferShape", ElemwiseShape<-1, 1>)
@@ -197,13 +197,13 @@ NNVM_REGISTER_OP(_backward_Activation)
.set_attr("FInplaceOption", [](const NodeAttrs& attrs){
return std::vector >{{0, 0}};
})
-#if MXNET_USE_MKLDNN == 1
+#if MXNET_USE_MKLDNN == 100
.set_attr("FResourceRequest", [](const NodeAttrs& n) {
return std::vector{ResourceRequest::kTempSpace};
})
#endif
.set_attr_parser(ParamParser)
-#if MXNET_USE_MKLDNN == 1
+#if MXNET_USE_MKLDNN == 100
.set_attr("TIsMKLDNN", true)
.set_attr("FComputeEx", ActivationGradComputeExCPU)
#endif
diff --git a/src/operator/nn/mkldnn/mkldnn_act-inl.h
b/src/operator/nn/mkldnn/mkldnn_act-inl.h
index 6bf30e3..57507a5 100644
--- a/src/operator/nn/mkldnn/mkldnn_act-inl.h
+++ b/src/operator/nn/mkldnn/mkldnn_act-inl.h
@@ -20,7 +20,7 @@
/*!
* Copyright (c) 2019 by Contributors
* \file mkldnn_act-inl.h
- * \brief MKLDNN(Quantized) Activation operator based on subgraph
+ * \brief MKLDNN Activation operator
* /author Zhiyuan Huang
*/
@@ -28,20 +28,17 @@
#define MXNET_OPERATOR_NN_MKLDNN_MKLDNN_ACT_INL_H_
-#if MXNET_USE_MKLDNN == 1
+#if MXNET_USE_MKLDNN == 100
#include
#include
#include "../activation-inl.h"
-#include "./mkldnn_ops-inl.h"
-#include "./mkldnn_base-inl.h"
namespace mxnet {
namespace op {
mkldnn::algorithm GetMKLDNNActAlgo(const ActivationParam& param);
mkldnn::eltwise_forward::primitive_desc GetActFwdDescImpl(
-const ActivationParam& param, bool is_train,
-const mkldnn::memory &input_mem, int dtype);
+const ActivationParam& param, bool is_train, const mkldnn::memory
&input_mem);
class MKLDNNActForward {
public:
@@ -49,14 +46,13 @@ class MKLDNNActForward {
MKLDNNActForward(const ActivationParam& param, bool is_train,
const NDArray &data, const mkldnn::memory &mem): fwd_pd(
- GetActFwdDescImpl(param, is_train, mem, data.dtype()))
{}
- void SetNewMem(const mkldnn::memory &data, const mkldnn::memory &output);
- const mkldnn::eltwise_forward &GetFwd() const;
+ GetActFwdD
[GitHub] [incubator-mxnet] TaoLv merged pull request #16195: [mkldnn-v1.0] Add MKL-DNN activation
TaoLv merged pull request #16195: [mkldnn-v1.0] Add MKL-DNN activation URL: https://github.com/apache/incubator-mxnet/pull/16195 This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-mxnet] TaoLv closed pull request #16213: [mkldnn-v1.0][Don't merge] Trigger CI after merging the master branch
TaoLv closed pull request #16213: [mkldnn-v1.0][Don't merge] Trigger CI after merging the master branch URL: https://github.com/apache/incubator-mxnet/pull/16213 This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-mxnet] ybz79 opened a new issue #16227: How to use a pre-trained model in model zoo as a sub-block in my own model?
ybz79 opened a new issue #16227: How to use a pre-trained model in model zoo as a sub-block in my own model? URL: https://github.com/apache/incubator-mxnet/issues/16227 ## Description I want to use a pretrained model in the model zoo (such as BERT) as a layer in my own model. ```python class Mymodel(nn.Block): def __init__(bert_model): self.bert = bert_model . bert_model = get_model() model = Mymodel(bert_model) ``` But I have a small problem about the naming, since bert_model and other parameters in my own model didn't have a same prefix, which can cause error in model.save_params() , how can I slove this problem? This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[incubator-mxnet-site] branch asf-site updated: Bump the publish timestamp.
This is an automated email from the ASF dual-hosted git repository. aaronmarkham pushed a commit to branch asf-site in repository https://gitbox.apache.org/repos/asf/incubator-mxnet-site.git The following commit(s) were added to refs/heads/asf-site by this push: new ca711dc Bump the publish timestamp. ca711dc is described below commit ca711dc7bd0251ecddf672b4a8a9215b8d34d79f Author: mxnet-ci AuthorDate: Fri Sep 20 13:13:44 2019 + Bump the publish timestamp. --- date.txt | 1 + 1 file changed, 1 insertion(+) diff --git a/date.txt b/date.txt new file mode 100644 index 000..6719902 --- /dev/null +++ b/date.txt @@ -0,0 +1 @@ +Fri Sep 20 13:13:44 UTC 2019
[GitHub] [incubator-mxnet] wkcn commented on issue #15678: [MXNET-1418]Add contrib op into cpp package
wkcn commented on issue #15678: [MXNET-1418]Add contrib op into cpp package URL: https://github.com/apache/incubator-mxnet/pull/15678#issuecomment-533546216 I don't know how to solve the issue about Scala binding. Does anyone help me address the problem? Thank you! This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-mxnet] wkcn opened a new pull request #16226: Fix wrong results of min([inf, inf]) and max([-inf, -inf])
wkcn opened a new pull request #16226: Fix wrong results of min([inf, inf]) and max([-inf,-inf]) URL: https://github.com/apache/incubator-mxnet/pull/16226 ## Description ## Hi, there. In the latest version of MXNet, the results of min([inf, inf]) and max([-inf, -inf]) are wrong. #16206 . The reason is that the initial values of minimum or maximum reducer are the MaxValue and the MinValue. They should be the +InfValue and the -InfValue. In this PR, I add `PosInfValue` and `NegInfValue` in mshadow. Then I fixed the bug. ## Checklist ## ### Essentials ### Please feel free to remove inapplicable items for your PR. - [ ] The PR title starts with [MXNET-$JIRA_ID], where $JIRA_ID refers to the relevant [JIRA issue](https://issues.apache.org/jira/projects/MXNET/issues) created (except PRs with tiny changes) - [ ] Changes are complete (i.e. I finished coding on this PR) - [ ] All changes have test coverage: - Unit tests are added for small changes to verify correctness (e.g. adding a new operator) - Nightly tests are added for complicated/long-running ones (e.g. changing distributed kvstore) - Build tests will be added for build configuration changes (e.g. adding a new build option with NCCL) - [ ] Code is well-documented: - For user-facing API changes, API doc string has been updated. - For new C++ functions in header files, their functionalities and arguments are documented. - For new examples, README.md is added to explain the what the example does, the source of the dataset, expected performance on test set and reference to the original paper if applicable - Check the API doc at http://mxnet-ci-doc.s3-accelerate.dualstack.amazonaws.com/PR-$PR_ID/$BUILD_ID/index.html - [ ] To the my best knowledge, examples are either not affected by this change, or have been fixed to be compatible with this change ### Changes ### - [ ] Feature1, tests, (and when applicable, API doc) - [ ] Feature2, tests, (and when applicable, API doc) ## Comments ## - If this change is a backward incompatible change, why must this change be made. - Interesting edge cases to note here This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-mxnet] pengzhao-intel commented on issue #16195: [mkldnn-v1.0] Add MKL-DNN activation
pengzhao-intel commented on issue #16195: [mkldnn-v1.0] Add MKL-DNN activation URL: https://github.com/apache/incubator-mxnet/pull/16195#issuecomment-533495711 @TaoLv please merge the related 1.0 PR when the branch is good. This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-mxnet] pengzhao-intel commented on issue #16075: Integrate MKL-DNN leakyrelu
pengzhao-intel commented on issue #16075: Integrate MKL-DNN leakyrelu URL: https://github.com/apache/incubator-mxnet/pull/16075#issuecomment-533495090 + @ZhennanQin This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-mxnet] xinyu-intel commented on issue #16075: Integrate MKL-DNN leakyrelu
xinyu-intel commented on issue #16075: Integrate MKL-DNN leakyrelu URL: https://github.com/apache/incubator-mxnet/pull/16075#issuecomment-533489831 @pengzhao-intel I'll start another PR to enable gelu after this PR merged:) This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-mxnet] matteosal edited a comment on issue #16143: Failure of MKL-DNN Convolution from C API
matteosal edited a comment on issue #16143: Failure of MKL-DNN Convolution from C API URL: https://github.com/apache/incubator-mxnet/issues/16143#issuecomment-533484928 Any news about this? Not sure it should be labeled C API anymore, because one 1 example out of 2 (the second above here) can also be reproduced from Python. This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-mxnet] matteosal commented on issue #16143: Failure of MKL-DNN Convolution from C API
matteosal commented on issue #16143: Failure of MKL-DNN Convolution from C API URL: https://github.com/apache/incubator-mxnet/issues/16143#issuecomment-533484928 Any news about this? Not sure it should be labeled C API anymore, because one 1 example out of 2 (the one above here) can also be reproduced from Python. This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-mxnet] gasgallo opened a new pull request #16225: Add support for UpSampling op in mx2onnx
gasgallo opened a new pull request #16225: Add support for UpSampling op in mx2onnx URL: https://github.com/apache/incubator-mxnet/pull/16225 This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-mxnet] Sumanth-393 opened a new issue #16224: mxnet.base.MXNetError: [13:43:35] C:\Jenkins\workspace\mxnet-tag\mxnet\src\ndarray\ndarray.cc:1755: Check failed: fi->Read(data) Inval
Sumanth-393 opened a new issue #16224: mxnet.base.MXNetError: [13:43:35]
C:\Jenkins\workspace\mxnet-tag\mxnet\src\ndarray\ndarray.cc:1755: Check failed:
fi->Read(data) Invalid NDArray file format while loading resnet-18
URL: https://github.com/apache/incubator-mxnet/issues/16224
got Error while loading checkpoint from model.load_checkpoint
## Steps to reproduce
(Paste the commands you ran that produced the error.)
path='http://data.mxnet.io/models/imagenet/'
[
mx.test_utils.download(path+'resnet/18-layers/resnet-18-.params'),
mx.test_utils.download(path+'resnet/18-layers/resnet-18-symbol.json'),
mx.test_utils.download(path+'synset.txt')]
ctx = mx.cpu()
sym, arg_params, aux_params = mx.model.load_checkpoint('resnet-18', 0)
error is as below:
E:\my_projects\mxnet_work>python Mxnet_imageClaify.py
[13:43:35]
C:\Jenkins\workspace\mxnet-tag\mxnet\src\nnvm\legacy_json_util.cc:209: Loading
symbol saved by previous version v0.8.0. Attempting to upgrade...
[13:43:35]
C:\Jenkins\workspace\mxnet-tag\mxnet\src\nnvm\legacy_json_util.cc:217: Symbol
successfully upgraded!
Traceback (most recent call last):
File "Mxnet_imageClaify.py", line 14, in
sym, arg_params, aux_params = mx.model.load_checkpoint('resnet-18', 0)
File
"C:\Users\bunny\AppData\Local\Programs\Python\Python36\lib\site-packages\mxnet\model.py",
line 439, in load_checkpoint
save_dict = nd.load('%s-%04d.params' % (prefix, epoch))
File
"C:\Users\bunny\AppData\Local\Programs\Python\Python36\lib\site-packages\mxnet\ndarray\utils.py",
line 175, in load
ctypes.byref(names)))
File
"C:\Users\bunny\AppData\Local\Programs\Python\Python36\lib\site-packages\mxnet\base.py",
line 252, in check_call
raise MXNetError(py_str(_LIB.MXGetLastError()))
mxnet.base.MXNetError: [13:43:35]
C:\Jenkins\workspace\mxnet-tag\mxnet\src\ndarray\ndarray.cc:1755: Check failed:
fi->Read(data) Invalid NDArray file format
kindly help i'm new to mxnet
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org
With regards,
Apache Git Services
[GitHub] [incubator-mxnet] JiaoPaner commented on issue #15632: Building MxNet with CPP_PACKAGE on Windows10 (2019-07-23)
JiaoPaner commented on issue #15632: Building MxNet with CPP_PACKAGE on Windows10 (2019-07-23) URL: https://github.com/apache/incubator-mxnet/issues/15632#issuecomment-533443667 @QueensGambit I read the #15144 ,modified the line " #if !defined(_MSC_VER)" which is 744 to "#if !(defined(_MSC_VER) && _MSC_VER < 1900)" .Then, compile again that config includes USE_CPP_PACKAGE, my operating system is win10 x64 and IDE is Visual Studio 2015. After that the compile result has no any errors and no any above errors. And the examples can run without error , the solution does work. This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-mxnet] JiaoPaner edited a comment on issue #15143: dmlc::type_name_helper specialization of mxnet::tuple should not be disabled for MSVC
JiaoPaner edited a comment on issue #15143: dmlc::type_name_helper specialization of mxnet::tuple should not be disabled for MSVC URL: https://github.com/apache/incubator-mxnet/issues/15143#issuecomment-533442504 I read the #15144 ,modified the line " #if !defined(_MSC_VER)" which is 744 to "#if !(defined(_MSC_VER) && _MSC_VER < 1900)" .Then, compile again that config includes USE_CPP_PACKAGE, my operating system is win10 x64 and IDE is Visual Studio 2015. After that the compile result has no any errors and no any above errors. And the examples can run without error , the solution does work. This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-mxnet] pengzhao-intel commented on issue #16075: Integrate MKL-DNN leakyrelu
pengzhao-intel commented on issue #16075: Integrate MKL-DNN leakyrelu URL: https://github.com/apache/incubator-mxnet/pull/16075#issuecomment-533443566 @xinyu-intel the MKL-DNN is updated to 0.21 now. This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[incubator-mxnet] branch master updated (7126438 -> 986cecd)
This is an automated email from the ASF dual-hosted git repository. patriczhao pushed a change to branch master in repository https://gitbox.apache.org/repos/asf/incubator-mxnet.git. from 7126438 New Website: New Pipeline [3/3] (#15883) add 986cecd Update MKL-DNN dependency (#16073) No new revisions were added by this update. Summary of changes: 3rdparty/mkldnn | 2 +- ci/docker/install/ubuntu_mklml.sh | 2 +- cmake/DownloadMKLML.cmake | 2 +- 3 files changed, 3 insertions(+), 3 deletions(-)
[GitHub] [incubator-mxnet] pengzhao-intel merged pull request #16073: Update MKL-DNN dependency
pengzhao-intel merged pull request #16073: Update MKL-DNN dependency URL: https://github.com/apache/incubator-mxnet/pull/16073 This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-mxnet] JiaoPaner commented on issue #15143: dmlc::type_name_helper specialization of mxnet::tuple should not be disabled for MSVC
JiaoPaner commented on issue #15143: dmlc::type_name_helper specialization of mxnet::tuple should not be disabled for MSVC URL: https://github.com/apache/incubator-mxnet/issues/15143#issuecomment-533442504 I read the #15144 ,modified the line " #if !defined(_MSC_VER)" which is 744 to "#if !(defined(_MSC_VER) && _MSC_VER < 1900)" .Then, compile again that config includes USE_CPP_PACKAGE.After that the compile result has no any errors and no any above errors. And the examples can run without error , the solution does work. This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-mxnet] rongzha1 opened a new pull request #16223: [mkldnn-v1.0] Add MKL-DNN LRN
rongzha1 opened a new pull request #16223: [mkldnn-v1.0] Add MKL-DNN LRN URL: https://github.com/apache/incubator-mxnet/pull/16223 ## Description ## (Brief description on what this PR is about) ## Checklist ## ### Essentials ### Please feel free to remove inapplicable items for your PR. - [ ] The PR title starts with [MXNET-$JIRA_ID], where $JIRA_ID refers to the relevant [JIRA issue](https://issues.apache.org/jira/projects/MXNET/issues) created (except PRs with tiny changes) - [ ] Changes are complete (i.e. I finished coding on this PR) - [ ] All changes have test coverage: - Unit tests are added for small changes to verify correctness (e.g. adding a new operator) - Nightly tests are added for complicated/long-running ones (e.g. changing distributed kvstore) - Build tests will be added for build configuration changes (e.g. adding a new build option with NCCL) - [ ] Code is well-documented: - For user-facing API changes, API doc string has been updated. - For new C++ functions in header files, their functionalities and arguments are documented. - For new examples, README.md is added to explain the what the example does, the source of the dataset, expected performance on test set and reference to the original paper if applicable - Check the API doc at http://mxnet-ci-doc.s3-accelerate.dualstack.amazonaws.com/PR-$PR_ID/$BUILD_ID/index.html - [ ] To the my best knowledge, examples are either not affected by this change, or have been fixed to be compatible with this change ### Changes ### - [ ] Feature1, tests, (and when applicable, API doc) - [ ] Feature2, tests, (and when applicable, API doc) ## Comments ## - If this change is a backward incompatible change, why must this change be made. - Interesting edge cases to note here add mkldnn lrn @TaoLv @ciyongch @ZhennanQin @PatricZhao This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-mxnet] xidulu edited a comment on issue #15857: [Numpy] Added operator logaddexp; added support for zero-size tensor in BinaryBroadcastBackwardUseIn
xidulu edited a comment on issue #15857: [Numpy] Added operator logaddexp; added support for zero-size tensor in BinaryBroadcastBackwardUseIn URL: https://github.com/apache/incubator-mxnet/pull/15857#issuecomment-533434271 @szha Conflicts resolved (for now) This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-mxnet] xidulu commented on issue #15857: [Numpy] Added operator logaddexp; added support for zero-size tensor in BinaryBroadcastBackwardUseIn
xidulu commented on issue #15857: [Numpy] Added operator logaddexp; added support for zero-size tensor in BinaryBroadcastBackwardUseIn URL: https://github.com/apache/incubator-mxnet/pull/15857#issuecomment-533434271 @szha Conflicts solved (for now) This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
